WO2009146998A1 - Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure - Google Patents

Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure Download PDF

Info

Publication number
WO2009146998A1
WO2009146998A1 PCT/EP2009/055663 EP2009055663W WO2009146998A1 WO 2009146998 A1 WO2009146998 A1 WO 2009146998A1 EP 2009055663 W EP2009055663 W EP 2009055663W WO 2009146998 A1 WO2009146998 A1 WO 2009146998A1
Authority
WO
WIPO (PCT)
Prior art keywords
frame
file
frames
decoding
bit stream
Prior art date
Application number
PCT/EP2009/055663
Other languages
French (fr)
Inventor
Sven Kordon
Peter Jax
Johannes Boehm
Original Assignee
Thomson Licensing
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Thomson Licensing filed Critical Thomson Licensing
Priority to JP2011510935A priority Critical patent/JP5774983B2/en
Priority to US12/737,007 priority patent/US8731946B2/en
Priority to EP09757381A priority patent/EP2289245B1/en
Priority to PL09757381T priority patent/PL2289245T3/en
Priority to CN200980120515.2A priority patent/CN102047336B/en
Priority to ES09757381T priority patent/ES2379321T3/en
Priority to AT09757381T priority patent/ATE545281T1/en
Publication of WO2009146998A1 publication Critical patent/WO2009146998A1/en

Links

Classifications

    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B20/00Signal processing not specific to the method of recording or reproducing; Circuits therefor
    • G11B20/10Digital recording or reproducing
    • G11B20/10527Audio or video recording; Data buffering arrangements
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/04Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
    • G10L19/16Vocoder architecture
    • G10L19/167Audio streaming, i.e. formatting and decoding of an encoded audio signal representation into a data stream for transmission or storage purposes
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/005Correction of errors induced by the transmission channel, if related to the coding algorithm
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B20/00Signal processing not specific to the method of recording or reproducing; Circuits therefor
    • G11B20/00086Circuits for prevention of unauthorised reproduction or copying, e.g. piracy
    • G11B20/00166Circuits for prevention of unauthorised reproduction or copying, e.g. piracy involving measures which result in a restriction to authorised contents recorded on or reproduced from a record carrier, e.g. music or software
    • G11B20/00181Circuits for prevention of unauthorised reproduction or copying, e.g. piracy involving measures which result in a restriction to authorised contents recorded on or reproduced from a record carrier, e.g. music or software using a content identifier, e.g. an international standard recording code [ISRC] or a digital object identifier [DOI]
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B20/00Signal processing not specific to the method of recording or reproducing; Circuits therefor
    • G11B20/10Digital recording or reproducing
    • G11B20/12Formatting, e.g. arrangement of data block or words on the record carriers
    • G11B20/1262Formatting, e.g. arrangement of data block or words on the record carriers with more than one format/standard, e.g. conversion from CD-audio format to R-DAT format
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B20/00Signal processing not specific to the method of recording or reproducing; Circuits therefor
    • G11B20/00007Time or data compression or expansion
    • G11B2020/00014Time or data compression or expansion the compressed signal being an audio signal
    • G11B2020/00021Time or data compression or expansion the compressed signal being an audio signal lossless audio compression
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11BINFORMATION STORAGE BASED ON RELATIVE MOVEMENT BETWEEN RECORD CARRIER AND TRANSDUCER
    • G11B20/00Signal processing not specific to the method of recording or reproducing; Circuits therefor
    • G11B20/00007Time or data compression or expansion
    • G11B2020/00014Time or data compression or expansion the compressed signal being an audio signal
    • G11B2020/00057MPEG-1 or MPEG-2 audio layer III [MP3]

Definitions

  • the invention relates to a frame-based data format with changeable format, the frames containing encoded signal data which for decoding of a current frame require access to pre ⁇ vious frame data.
  • MPEGl Audio Layer III bit stream with its bit reservoir technique, where data for a current frame is stored in previous frames. The position of the data is indicated in the bit stream by a pointer to the position in byte of the beginning of the main-data in the previous frame. If the decoder did not receive the required previous frame, decoding of the current mp3 frame is skipped.
  • the MPEGl Audio Layer III bit stream does not have a bit stream header field for enabling sample-exact cutting.
  • the present invention can be applied for a data format or structure where an extension data stream provides lossless extension for a lossy base layer data stream for e.g. an audio signal, e.g. the hd3 audio file format for lossless ex ⁇ tension of an mp3 bit stream, wherein decoding of a frame requires decoding of more than one previous frame and the bit stream format offers an archival mode and a streaming mode, as well as a sample-exact cutting feature of the bit stream .
  • an extension data stream provides lossless extension for a lossy base layer data stream for e.g. an audio signal, e.g. the hd3 audio file format for lossless ex ⁇ tension of an mp3 bit stream, wherein decoding of a frame requires decoding of more than one previous frame and the bit stream format offers an archival mode and a streaming mode, as well as a sample-exact cutting feature of the bit
  • ⁇ Archival mode' means that a file includes a single bit stream header and successive frames of encoded or decoded signal samples or coefficients (in particular audio sam- pies) , whereby decoding must start at the beginning of the file, i.e. with the frame following the bit stream header.
  • ⁇ Streaming mode' means that a coherent bit stream is sepa ⁇ rated into more than one package of data, each package starting with a single bit stream header followed by several frames, whereby decoding can start at each package.
  • Sample-exact cutting means generating as a new bit stream a partial section from an existing bit stream, thereby encoding or decoding only a subset of samples of that existing bit stream, wherein the first and the last decodable sample can be chosen arbitrarily.
  • the problem to be solved by the invention is to disclose a bit stream format that facilitates the initialisation of the decoder states at the beginning of a cut file or a streaming package. Depending on the frame, the number of previous frames involved in the decoder states recovery is different. A streaming mode file or a cut file can start at each frame and therefore the required number of previous frames for the decoder state recovery is missing.
  • This problem is solved by the methods disclosed in claims 1 and 3 and by the data structure disclosed in claim 2. An ap ⁇ paratus that utilises the method of claim 3 is disclosed in claim 4.
  • the inventive bit stream format can be generated or used at encoder side as well as at decoder side.
  • the invention is related to a corresponding decoder processing and the required bit stream header fields for initialising the decoder states at the beginning of the cut file or the stream package.
  • the bit stream it is signalised in the bit stream how many frames are required for initialising the decoder states. This is performed explicitly by signalising within the bit stream header the number of frames required for the decoder status recovery. These frames are used for decoder initialisation but not for decoding samples or coefficients. For the sample-exact cutting, for the frame at which the cut shall occur, the number of samples or coeffi ⁇ cients to be muted is indicated in the bit stream header, for enabling a sample-exact cutting beyond the frame bor- ders .
  • the invention can be used e.g. in an hd3 encoder in order to allow two bit stream modes and sample-exact cutting of the bit stream. It is a specific solution because decoding of a frame of the second layer of the hd3 encoder bit stream re ⁇ quires more then one decoded frame of the first layer, whereas in most known frame-based audio formats each frame can be decoded independently from the other frames. Coders that make use of backward adaptive prediction tech- niques can not use the inventive processing because it re ⁇ quires that all previously decoded samples are known for re ⁇ covering the prediction states. Therefore such known bit streams include the prediction states for initialisation of the decoder after a defined number of coded samples, which is not true for the present invention.
  • the inventive method is suited for generating a frame based bit stream format file including at least one bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein in said header in ⁇ formation items are arranged representing:
  • the number of initial frames to mute while decoding states are initialised, and optionally, for a sample-exact cutting feature, the number of samples to mute in the corresponding frame;
  • said file includes a single bit stream header section and successive ones of said frames, and a decoding of said encoded signal data is to start with the first one of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute signalises how many ones of initial frames in a current super frame are required for establish ⁇ ing data for initialising states of a decoding of said signal data before actually starting decoding of encoded signal data from the following frame or frames of said current su- per frame, such initial frames not being used for decoding the encoded signal data contained therein.
  • the inventive data structure is suited for a frame based bit stream format file including at least one bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein in said header in ⁇ formation items are arranged representing: - total number of samples per channel for said file;
  • the number of initial frames to mute while decoding states are initialised, and optionally, for a sample-exact cutting feature, the number of samples to mute in the corresponding frame;
  • said file includes a single bit stream header section and successive ones of said frames, and a decoding of said encoded signal data is to start with the first one of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute signalises how many ones of initial frames in a current super frame are required for establish ⁇ ing data for initialising states of a decoding of said signal data before actually starting decoding of encoded signal data from the following frame or frames of said current su ⁇ per frame, such initial frames not being for use for decod ⁇ ing the encoded signal data contained therein.
  • one inventive method is suited for cutting a section out of a frame based bit stream format file includ ⁇ ing a bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header in- eludes information items representing:
  • a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said method including the step of forming a cutting file by taking the required frames data from said bit stream and ar ⁇ ranging a cutting header in front of these frames data, wherein said cutting header is derived from said header and in said cutting header:
  • said information item regarding decoding of said encoded signal data is to start at the first frame of said file is changed to decoding of said encoded signal data is to start later in said file; - an information item is added regarding the number of initial frames to mute while decoding states are initialised.
  • one inventive apparatus is suited for cutting a section out of a frame based bit stream format file includ- ing a bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header in ⁇ cludes information items representing: - total number of samples per channel for said file; an 'archival mode' for said file; a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said apparatus including means being adapted for forming a cutting file by taking the required frames data from said bit stream and arranging a cutting header in front of these frames data, wherein said cutting header is derived from said header, and in said cutting header:
  • one inventive method is suited for changing a frame based bit stream file format, said bit stream includ ⁇ ing a header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header includes informa ⁇ tion items representing:
  • a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said method including the step of generating from said archival mode file a 'streaming mode' bit stream by forming from every successive group of frames of said archival mode file a super frame of said streaming mode bit stream by tak- ing the required frame data from said archival mode file and arranging a super frame header at the beginning of each su- per frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers :
  • an information item is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is '0';
  • one inventive apparatus is suited for changing a frame based bit stream file format, said bit stream in ⁇ cluding a header section, said frames including encoded signal data, wherein data required for a decoding or an evalua ⁇ tion of data of a current frame can be contained in one or more previous frames, and wherein said header includes in ⁇ formation items representing: - total number of samples per channel for said file;
  • a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said apparatus including means being adapted for generating from said archival mode file a 'streaming mode' bit stream by forming from every successive group of frames of said ar ⁇ chival mode file a super frame of said streaming mode bit stream by taking the required frame data from said archival mode file and arranging a super frame header at the beginning of each super frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers :
  • an information item is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is
  • the last-mentioned method can be used for de ⁇ coding said streaming mode bit stream, further including the steps of: - when starting decoding of said streaming mode bit stream at said first super frame, the decoding of the encoded signal data starts at the first frame of that super frame using default decoder states;
  • a number x muteFrames' of frames corresponding to said information item regarding the number of initial frames to mute is used for initialising the decoding states, and the decoding of encoded signal data starts at frame number muteFrames+1 of the current super frame using these decoding states;
  • the following description provides the corresponding bit stream information and the decoder processing for having one bit stream format that supports a streaming mode as well as an archival mode and that facilitates sample-exact cutting.
  • FIG. 1 Structure of an archival mode bit stream
  • FIG. 2 Structure of a cut archival mode bit stream
  • Fig. 4 Dependencies of the n-th frame, whereby decoding of frame n _4 requires data stored in a previous super frame, and therefore it is invalid for stand-alone decoding of the super frame.
  • Frame n _3, frame n _2 and frame n _]_ are required for initialising the decoder states of frame n ;
  • Fig. 5 Decoder flow chart for the processing of archival mode, cutting mode and streaming mode of the inventive bit stream format;
  • Fig. 6 Simplified HD3 file format.
  • the invention was made during the hd3 codec development. Therefore, the description is applicable to the current ver- sion of the hd3 bit stream format. However, the invention can be applied correspondingly to all frame based file for- mats that require knowledge of a number of previous frames for decoding a frame and wherein number is written into the bit stream header for initialising the decoder states, or for achieving sample-exact cutting by indicating the number of samples to mute of the decoded frame.
  • the hd3 codec processing enables storage, in a single file, of an mp3 file plus extension data for the bit-exact repre ⁇ sentation of the source of the mp3 file.
  • the first version of the hd3 bit stream format was an archival format with one file header and successive data frames Frame ⁇ , Frame]_, ... Frame n , as shown in Fig. 1.
  • the different layers of the hd3 codec are not depicted here, and it is assumed that the described coder requires a number of previously decoded frames to decode the current frame.
  • the archival mode bit stream as shown in Fig. 1 consists of one file header at the beginning of the file, followed by signal data divided into frames, wherein each frame repre ⁇ sents code representing a segment of L coded signal samples or coefficients.
  • the archival mode bit stream decoding can start only at the first frame because only there the decoder states are known (they are defined in every decoding standard) .
  • the be ⁇ low-mentioned Main_Data_Begin_Pointer in mp3 is set to zero for the first frame.
  • the decoder states get initialised and a correct de- coding of the samples is obtained.
  • a main characteristic of the described format is that not all but only k, with 0 ⁇ k ⁇ K, previously decoded frames are required to achieve correct results, wherein K is the maximal number of required decoded frames and K is a constant value that is known in the en- coder and in the decoder.
  • K is the maximal number of required decoded frames
  • K is a constant value that is known in the en- coder and in the decoder.
  • decoding of the required information from the previous frames without knowledge of other frames needs to be feasible.
  • the hd3 codec provides such feature by the independently de- codable mp3 frames.
  • the reconstruction of the bit- exact replica i.e. lossless reconstruction
  • the MPEGl Audio Layer III standard ISO/IEC ISO/IEC
  • 11172-3 includes the above-mentioned bit reservoir technique that allows storage of main data in previous frames.
  • a main data begin pointer is used that points to the beginning of the main data located in a previous frame.
  • decoding of single mp3 frames requires information from previous frames as well.
  • the frames n-3 to n-1 can not be decoded cor ⁇ rectly because their decoder states are not initialised cor ⁇ rectly due to the missing data from the previous frames.
  • the correct decoder states of the following frame can be reconstructed in a decoder states initialisa ⁇ tion step/stage DSI by performing the mapping process de ⁇ scribed in EP08102308.7.
  • the first frame with correctly ini ⁇ tialised decoder states is Framen '
  • some extra information items are added to the bit stream header depicted in the following table. Based on these additional information items the decoder can decide if the processing will start with known default decoder states or with unknown decoder states:
  • Table 1 Section of the header of the inventive bit stream format
  • the x MuteIfFirstSuperFrame' field of the bit stream header indicates whether or not decoding of the following frames requires an additional initialisation of the decoder states, If true (i.e. the flag is set) , further information items are inserted into the bit stream header, which will be dis ⁇ cussed in the following examples for cutting and streaming.
  • the bit stream header for the archival mode uses the X OFI/ , 'FileMode' and 'MutelfFirstSuperFrame' data fields only, wherein the values x FileMode' and x MuteIfFirstSuperFrame' are set to zero (or the corresponding flags are cleared) to indicate a non-cut archival file.
  • Cutting is used at decoding side to separate a short sub ⁇ section from a received or replayed complete coherent bit stream in archival mode.
  • the bit stream mode of the cut file is equal to that of the archival mode format.
  • Fig. 2 depicts a cut file produced from the archival mode file format shown in Fig. 1. Cutting can be used, for example, to get a short preview of the complete file. Instead of decoding and newly encoding the desired section, the required frames are just cut out from the input file and a new header is inserted in front of these cut-out data frames.
  • the problem with cutting is that for the decoding of the first frame of the cut-out data frames the data from the re ⁇ quired previous frames are missing in case the preview or the cut-out section, respectively, does not start at the be- ginning of the complete bit stream. Therefore the cut file evaluates the x MuteIfFirstSuperFrame' data field of the bit stream header indicating that the processing of the first frames merely initialises the decoder states and that the decoded samples of these frames are to be muted.
  • the number of frames that are to be muted is indicated in the bit stream header in the x muteFrames' field, and the number x muteSamples' of samples that are to be muted is indicated in the header to enable sample-exact cutting instead of pure frame-exact cutting.
  • OFL orig Total length of the input file in samples per channel
  • OFL cut Total length of the cut file in samples per chan- nel
  • Frame va i ld First decodable frame with known decoder states
  • Frame st art First decoded frame of the cut file Create the cut file opy the header followed by the frames Frame sta rt to Frame en d
  • x FileMode' is set to x 0' and 'MutelfFirstSuperFrame' is set to ⁇ l' .
  • the cut file must start with the frame that holds the data of the first frame involved in the decoder state recovery of x Frame va i ld ' .
  • the obtained number of previous frames is written into the 'muteFrame' field of the bit stream header, such that the decoder knows that these frames are to be used only for ini ⁇ tialisation of the decoder states and not for decoding samples .
  • an additional number of 'samples per channel to mute' can be indi ⁇ cated in the header as well. These samples will be decoded correctly by the decoder but will not be presented to the user. Therefore the decoded presented signal can start with a sample located anywhere within the frame, instead of starting at the beginning of the frame.
  • X OFI/ is the number of samples per channel that can be decoded with known decoder states. This is ex ⁇ plained in more detail in the following Streaming bit stream mode section. Therefore the samples of the initialisation frames are to be added to the actual number of samples of the cut file. The decoder will automatically subtract these additional samples from the X OFL' value in the decoder proc ⁇ ess. Further details are described in the Decoding process ⁇ ing section.
  • the archival bit stream mode can be transformed at decoding side into a streaming bit stream mode.
  • 'Streaming mode' means that the frames of the archival bit stream are separated into successive packages, whereby each one of these packages is called a 'super frame' (SF) and has the same structure as an archival bit stream.
  • SF 'super frame'
  • the x FileMode' data field of the bit stream header of each super frame carries the value x l' indicating the streaming mode, in which the decoder must decode several successive super frames in order to reconstruct all samples of the en ⁇ coded file.
  • the first super frame of a streaming mode stream is basically identical to the beginning of an archival mode file, but in the bit stream header the data fields xFileMode' and X OFL' carry values different than in an ar ⁇ chival mode header.
  • the decoder starts decoding of the first super frame, the default decoder states are used and the first frame can be decoded directly without requiring any further information items.
  • a streaming mode bit stream can be replayed or decoded starting at each one of its super frames.
  • AND decoding starts at the beginning of a new coherent bit stream in streaming mode.
  • the default decoder states are used and the decoding of the samples can start directly at the first frame of the super frame .
  • the first 'muteFrame' frames of the current super frame are used for initialising the decoder states, and the decoding of samples can start at frame number ('muteFrames' +1) .
  • the x StreamingCheckSum' data field of the header is used. Its value is obtained from the data of the previous super frame and is written into the 'Stream- ingCheckSum' data field of the following super frame, so that successive super frames can be identified. For example, a Cyclic Redundancy Check sum (CRC) or a hash value of frames, e.g. the last frame, of the previous super frame can be used.
  • CRC Cyclic Redundancy Check sum
  • the decoder computes the CRC of the previous super frame and compares it with the value stored in the header of the current super frame.
  • the decoder states are to be re ⁇ initialised using the first x muteFrames' number of frames of the next super frame. Decoding of samples starts at frame number ('muteFrames' +1) .
  • the x StreamingCheckSum' data field is not used in the first super frame of a coherent streaming mode bit stream because previous frames are not required and therefore the consis ⁇ tency check is superfluous.
  • the decoder status of the previously decoded super frame can be used to decode the frames of the next super frame .
  • a typical streaming processing starts with an initialisation phase followed by several decoding phases.
  • the 'muteFrame' information is not used because the decoder states are initialised correctly by corresponding data from the previous super frame, and the data to which the x Main_Data_Begin_Pointer' is pointing is available in the previous super frame.
  • OFL or -j_g Total length of the input file in samples per chan ⁇ nel M: Number of super frames with M ⁇ ceil (OFL or -j_g/L)
  • M mean : Average number of frames per super frame
  • NumFrames ceil (OFL or -j_g/L) , wherein x ceil' returns a value representing the smallest integer that is greater equal the argument of the ceil function.
  • Copy frames number x 0' to (M mean -1) Copy frames number x 0' to (M mean -1) .
  • Compute x StreamingCheckSum' compute the Cyclic Redundancy Check sum of a defined part of the previous super frame and store the result in data field ⁇ StreamingCheckSuir ⁇ p' .
  • Step/stage 1 the input file or bit stream is available for reading the required data.
  • Step/stage 2 finds and reads the first header of the file or bit stream, i.e. the header information items of Table 1 are set, stored or loaded.
  • Step/stage 3 sets the variable x Sam- plesToMute' to its default value zero.
  • X I/ is the number of samples that are decoded from one frame, which is a known decoder constant.
  • Step/stage 6 decodes all received frames which includes also the first x muteFrames' frames. Hence X OFI/ decoded samples are passed to the next step/stage.
  • the first x SamplesToMute' samples are invalid, they are used for initialising the de ⁇ coder states.
  • step/stage 7 removes the invalid x Sam- plesToMute' samples and returns only the remaining samples to step/stage 8.
  • the number of remaining samples is the dif ⁇ ference between the value x SamplesToMute' and the total num ⁇ ber OFL of decoded samples. In case that x SamplesToMute' is greater than OFL, zero samples are returned to step/stage 8. This can happen in the streaming-mode if the number of frames per super frame is less than the number of frames to mute .
  • MuteNextSF SamplesToMute - OFL is calculated in step/stage 9 in order to store the number of samples to mute in the next super frame .
  • step/stage 10 the decoding processing is finished 11 for an archival mode file or bit stream or a cut file bit stream because all OFL samples of the file have been decoded.
  • step/stage 10 checks the x FileMode' and proceeds to END step 11 for stopping the archival mode decoding process. Otherwise, decoding of the streaming-mode bit stream is to be continued. Therefore the x StreamingCheckSum' , for example a CRC, is computed from the frame data of the currently decoded frame in step/stage 12. The processing and the data used for the x StreamingCheckSum' computation must produce identical results in the encoder and the decoder. Furthermore, the x StreamingCheckSum' should represent a clear identification of the current frame because it is used for verifying the consistency of the decoder states of the next super frame. Therefore the used data should diversify from super frame to super frame and represent the coded data of the super frame.
  • the x StreamingCheckSum' for example a CRC
  • step/stage 13 Upon switching to the next super frame, step/stage 13 searches and reads the header of the next super frame. This step/stage re-initialises all bit stream header variables so that the former header information is lost. Upon reaching the end of the file, or in case no valid bit stream header has been found, the decoder will proceed to step 11 and stop the decoding.
  • a streaming mode bit stream file can consists of two or more successive coherent bit streams. Therefore it is to be checked whether the following super frame is the first super frame of a new coherent bit stream, because in such case the current decoder states belong to a different bit stream and are to be reset to the default values.
  • Step/stage 15 checks variable x FirstSuperFrame' whether a new coherent bit stream is starting. If true, step/stage 17 resets the decoder states and starts decoding of the next super frame at step/stage 3. This is identical to starting decoding of a new file, and is also identical to the decoder initialisation phase 1 of the streaming-mode section in this description. Otherwise the next super frame belongs to the same coherent bit stream and the processing continues in step/stage 16.
  • the x StreamingCheckSum' value of the current super frame bit stream header is compared with the x Streaming-
  • step/stage 17 is used to reset the decoder states and then decoding of the next super frame starts at step/stage 3. Therefore the processing is in the decoder initialisation phase 3 of the streaming-mode section in this description and the decoding is started with default decoder states, which leads to the decoder initialisation phase 2 in step/stage 4.
  • step/stage 18 which sets the x MuteIfFirstSuper- Frame' variable to zero. Therefore step/stage 4 will not lead to step/stage 8 and the x muteFrames' value of the cur ⁇ rent super frame will not be used.
  • decoding of the current super frame must start at step/stage 4 in order to omit the re-initialisation of x SamplesToMute' in step/stage 3. If in step/stage 19 parameter MuteNextSF ⁇ 0, there are no remaining samples to mute and the decoder states are correct and the decoding of the current super frame can start di ⁇ rectly with the first frame. The number of samples to mute is set to zero in step/stage 3. In Streaming mode the decoding of successive super frames is repeated until step/stage 13 does no more find a bit stream header, and the decoding processing is stopped in step 11.
  • the invention facilitates processing of a frame based bit stream format in archival mode bit stream or streaming mode bit stream, and enables a sample-exact cutting of the archival mode bit stream, even if decoding of one frame re ⁇ quires information from previous frames.
  • inventive decoding processing is used in a hd3 decoder implementation and is tested successfully.
  • a HD3 file in it simplest format has three mandatory data sections shown in Fig. 6, and up to four optional data sec ⁇ tions.
  • the mandatory hd3lD data section provides file-mode, length, and CRC information.
  • This hd3lD section is encapsulated in an ancillary data section (silence PCM) of an mp3 frame.
  • Such ancillary data section of an mp3 frame may also contain a Xing- or VBRi-compliant variable bit rate header in front of the hd3lD information.
  • the mandatory mp3-data section encapsulates the mp3 data stream or parts thereof.
  • the synchronisation words of the mp3 stream may be scrambled.
  • the mandatory cd-data section provides the lossless exten ⁇ sion to the mp3 data for reconstructing a mathematically lossless copy of the original music with one or two audio channels having a temporal resolution corresponding to 32 kHz, 44.1 kHz or 48 kHz and an amplitude resolution of 16 bit/sample.
  • An unsynchronisation scheme can be used that avoids the occurrence of mp3 sync headers.
  • An optional hd-data section (not depicted) enables lossless reconstruction of high-definition studio formats with up to 24 bit/sample and sampling rates of 192 kHz.
  • the optional ID3/ID3v2 data section stores metadata relating to the embedded mp3 encoded and the lossless reconstruction encoded music. There may also be an optional ID3vl tag at the very end of the HD3 file (not depicted) .
  • the invention can also be used for other codecs or bit streams like AAC, and in mp3 where the additional data (e.g. the headers) are arranged in mp3 ancillary data fields, which a corresponding mp3 decoder can evaluate.
  • additional data e.g. the headers
  • mp3 ancillary data fields which a corresponding mp3 decoder can evaluate.

Abstract

In frame-based bit stream formats the data required for decoding a current frame are usually stored within the data section for that frame. One exception is the mp3 bit stream where data for a current frame is stored in previous frames. If the decoder did not receive the required previous frame, decoding of the current mp3 frame is skipped. The invention can be applied for such bit streams, in an archival mode, a streaming mode and a sample-exact cutting of an archival mode. In the streaming and cutting modes, new headers are established. The number of frames required for initialising the decoder status is signalised in the header, as well as a consistency check value in the streaming mode. These frames are used for decoder initialisation but not for decoding samples or coefficients. For a sample-exact cutting, for the frame at which the cut shall occur, the number of samples or coefficients to be muted is also indicated in the header. The invention can be applied for the hd3 audio file format for lossless extension of an mp3 bit stream.

Description

Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
The invention relates to a frame-based data format with changeable format, the frames containing encoded signal data which for decoding of a current frame require access to pre¬ vious frame data.
Background
In frame-based bit stream formats the data required for de¬ coding a current frame are usually stored within the data section for that frame.
An exception is the MPEGl Audio Layer III bit stream with its bit reservoir technique, where data for a current frame is stored in previous frames. The position of the data is indicated in the bit stream by a pointer to the position in byte of the beginning of the main-data in the previous frame. If the decoder did not receive the required previous frame, decoding of the current mp3 frame is skipped.
Invention
The MPEGl Audio Layer III bit stream does not have a bit stream header field for enabling sample-exact cutting. The present invention can be applied for a data format or structure where an extension data stream provides lossless extension for a lossy base layer data stream for e.g. an audio signal, e.g. the hd3 audio file format for lossless ex¬ tension of an mp3 bit stream, wherein decoding of a frame requires decoding of more than one previous frame and the bit stream format offers an archival mode and a streaming mode, as well as a sample-exact cutting feature of the bit stream .
^Archival mode' means that a file includes a single bit stream header and successive frames of encoded or decoded signal samples or coefficients (in particular audio sam- pies) , whereby decoding must start at the beginning of the file, i.e. with the frame following the bit stream header. ^Streaming mode' means that a coherent bit stream is sepa¬ rated into more than one package of data, each package starting with a single bit stream header followed by several frames, whereby decoding can start at each package.
Sample-exact cutting means generating as a new bit stream a partial section from an existing bit stream, thereby encoding or decoding only a subset of samples of that existing bit stream, wherein the first and the last decodable sample can be chosen arbitrarily.
The problem to be solved by the invention is to disclose a bit stream format that facilitates the initialisation of the decoder states at the beginning of a cut file or a streaming package. Depending on the frame, the number of previous frames involved in the decoder states recovery is different. A streaming mode file or a cut file can start at each frame and therefore the required number of previous frames for the decoder state recovery is missing. This problem is solved by the methods disclosed in claims 1 and 3 and by the data structure disclosed in claim 2. An ap¬ paratus that utilises the method of claim 3 is disclosed in claim 4.
The inventive bit stream format can be generated or used at encoder side as well as at decoder side. Regarding decoder side, the invention is related to a corresponding decoder processing and the required bit stream header fields for initialising the decoder states at the beginning of the cut file or the stream package.
According to the invention, it is signalised in the bit stream how many frames are required for initialising the decoder states. This is performed explicitly by signalising within the bit stream header the number of frames required for the decoder status recovery. These frames are used for decoder initialisation but not for decoding samples or coefficients. For the sample-exact cutting, for the frame at which the cut shall occur, the number of samples or coeffi¬ cients to be muted is indicated in the bit stream header, for enabling a sample-exact cutting beyond the frame bor- ders .
The invention can be used e.g. in an hd3 encoder in order to allow two bit stream modes and sample-exact cutting of the bit stream. It is a specific solution because decoding of a frame of the second layer of the hd3 encoder bit stream re¬ quires more then one decoded frame of the first layer, whereas in most known frame-based audio formats each frame can be decoded independently from the other frames. Coders that make use of backward adaptive prediction tech- niques can not use the inventive processing because it re¬ quires that all previously decoded samples are known for re¬ covering the prediction states. Therefore such known bit streams include the prediction states for initialisation of the decoder after a defined number of coded samples, which is not true for the present invention.
In principle, the inventive method is suited for generating a frame based bit stream format file including at least one bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein in said header in¬ formation items are arranged representing:
- total number of samples per channel for said file; - an 'archival mode' or a 'streaming mode' for said file;
- whether or not a decoding of said encoded signal data can start at the first frame of said file;
- if a decoding of said encoded signal data is not to start at said first frame of said file, the number of initial frames to mute while decoding states are initialised, and optionally, for a sample-exact cutting feature, the number of samples to mute in the corresponding frame;
- in said streaming mode, a value for identifying a previous super frame for a consistency check, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, and a decoding of said encoded signal data is to start with the first one of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute signalises how many ones of initial frames in a current super frame are required for establish¬ ing data for initialising states of a decoding of said signal data before actually starting decoding of encoded signal data from the following frame or frames of said current su- per frame, such initial frames not being used for decoding the encoded signal data contained therein.
In principle, the inventive data structure is suited for a frame based bit stream format file including at least one bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein in said header in¬ formation items are arranged representing: - total number of samples per channel for said file;
- an 'archival mode' or a 'streaming mode' for said file; - whether or not a decoding of said encoded signal data can start at the first frame of said file;
- if a decoding of said encoded signal data is not to start at said first frame of said file, the number of initial frames to mute while decoding states are initialised, and optionally, for a sample-exact cutting feature, the number of samples to mute in the corresponding frame;
- in said streaming mode, a value for identifying a previous super frame for a consistency check, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, and a decoding of said encoded signal data is to start with the first one of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute signalises how many ones of initial frames in a current super frame are required for establish¬ ing data for initialising states of a decoding of said signal data before actually starting decoding of encoded signal data from the following frame or frames of said current su¬ per frame, such initial frames not being for use for decod¬ ing the encoded signal data contained therein.
In principle one inventive method is suited for cutting a section out of a frame based bit stream format file includ¬ ing a bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header in- eludes information items representing:
- total number of samples per channel for said file; - an 'archival mode' for said file;
- a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said method including the step of forming a cutting file by taking the required frames data from said bit stream and ar¬ ranging a cutting header in front of these frames data, wherein said cutting header is derived from said header and in said cutting header:
- said total number of samples per channel for said file is replaced by the number of samples per channel for said cutting file; - said archival mode is kept;
- said information item regarding decoding of said encoded signal data is to start at the first frame of said file is changed to decoding of said encoded signal data is to start later in said file; - an information item is added regarding the number of initial frames to mute while decoding states are initialised.
In principle one inventive apparatus is suited for cutting a section out of a frame based bit stream format file includ- ing a bit stream header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header in¬ cludes information items representing: - total number of samples per channel for said file; an 'archival mode' for said file; a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said apparatus including means being adapted for forming a cutting file by taking the required frames data from said bit stream and arranging a cutting header in front of these frames data, wherein said cutting header is derived from said header, and in said cutting header:
- said total number of samples per channel for said file is replaced by the number of samples per channel for said cutting file;
- said archival mode is kept; - said information item regarding decoding of said encoded signal data is to start at the first frame of said file is changed to decoding of said encoded signal data is to start later in said file;
- an information item is added regarding the number of ini- tial frames to mute while decoding states are initialised.
In principle one inventive method is suited for changing a frame based bit stream file format, said bit stream includ¬ ing a header section, said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame can be contained in one or more previous frames, and wherein said header includes informa¬ tion items representing:
- total number of samples per channel for said file; - an 'archival mode' for said file;
- a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said method including the step of generating from said archival mode file a 'streaming mode' bit stream by forming from every successive group of frames of said archival mode file a super frame of said streaming mode bit stream by tak- ing the required frame data from said archival mode file and arranging a super frame header at the beginning of each su- per frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers :
- said total number of samples per channel for said file is replaced by a first adapted number of samples per channel;
- an information item for 'streaming mode' instead of 'archival mode' is given;
- said information item regarding decoding of said encoded signal data is to start at the first frame of said file is kept, and in the following ones of said super frame headers:
- said total number of samples per channel for said file is replaced by a second adapted number of samples per chan¬ nel; - said information item for 'archival mode' is replaced by an information item for 'streaming mode' ;
- said information item regarding decoding of said encoded signal data is to start at the first frame of said file is replaced by an information item regarding decoding of said encoded signal data is not to start at said first frame;
- an information item is added regarding the number of initial frames to mute while decoding states are initialised;
- optionally, an information item is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is '0';
- an information item is added regarding consistency check data, the value of which is derived from the data of the previous super frame.
In principle one inventive apparatus is suited for changing a frame based bit stream file format, said bit stream in¬ cluding a header section, said frames including encoded signal data, wherein data required for a decoding or an evalua¬ tion of data of a current frame can be contained in one or more previous frames, and wherein said header includes in¬ formation items representing: - total number of samples per channel for said file;
- an 'archival mode' for said file;
- a decoding of said encoded signal data is to start at the first frame of said file, wherein in said archival mode, said file includes a single bit stream header section and successive ones of said frames, said apparatus including means being adapted for generating from said archival mode file a 'streaming mode' bit stream by forming from every successive group of frames of said ar¬ chival mode file a super frame of said streaming mode bit stream by taking the required frame data from said archival mode file and arranging a super frame header at the beginning of each super frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers :
- said total number of samples per channel for said file is replaced by a first adapted number of samples per channel;
- an information item for 'streaming mode' instead of 'ar- chival mode' is given;
- said information item regarding decoding of said encoded signal data is to start at the first frame of said file is kept, and in the following ones of said super frame headers: - said total number of samples per channel for said file is replaced by a second adapted number of samples per chan¬ nel;
- said information item for 'archival mode' is replaced by an information item for 'streaming mode' ; - said information item regarding decoding of said encoded signal data is to start at the first frame of said file is replaced by an information item regarding decoding of said encoded signal data is not to start at said first frame;
- an information item is added regarding the number of ini- tial frames to mute while decoding states are initialised;
- optionally, an information item is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is
Figure imgf000012_0001
- an information item is added regarding consistency check data, the value of which is derived from the data of the previous super frame.
In principle, the last-mentioned method can be used for de¬ coding said streaming mode bit stream, further including the steps of: - when starting decoding of said streaming mode bit stream at said first super frame, the decoding of the encoded signal data starts at the first frame of that super frame using default decoder states;
- when not starting decoding of said streaming mode bit stream at said first super frame, following a decoding initialisation or reset, a number xmuteFrames' of frames corresponding to said information item regarding the number of initial frames to mute is used for initialising the decoding states, and the decoding of encoded signal data starts at frame number muteFrames+1 of the current super frame using these decoding states;
- when not starting decoding of said streaming mode bit stream at said first super frame, and consistency check data that were calculated from the previous super frame data is not consistent with corresponding consistency check data calculated from the current super frame, using a number xmuteFrames' of frames of the following super frame for re-initialising the decoding states starting decoding of encoded signal data at frame number xmuteFrames' +1 of that following super frame using these decoding states;
- when not starting decoding of said streaming mode bit stream at said first super frame, and a super frame was decoded before and said consistency check data are valid, using the decoder states of the previously decoded super frame to decode the encoded signal data of the frames of the current super frame.
Advantageous additional embodiments of the invention are disclosed in the respective dependent claims.
The following description provides the corresponding bit stream information and the decoder processing for having one bit stream format that supports a streaming mode as well as an archival mode and that facilitates sample-exact cutting.
Drawings
Exemplary embodiments of the invention are described with reference to the accompanying drawings, which show in:
Fig. 1 Structure of an archival mode bit stream;
Fig. 2 Structure of a cut archival mode bit stream;
Fig. 3 Structure of the streaming mode bit stream;
Fig. 4 Dependencies of the n-th frame, whereby decoding of framen_4 requires data stored in a previous super frame, and therefore it is invalid for stand-alone decoding of the super frame. Framen_3, framen_2 and framen_]_ are required for initialising the decoder states of framen; Fig. 5 Decoder flow chart for the processing of archival mode, cutting mode and streaming mode of the inventive bit stream format;
Fig. 6 Simplified HD3 file format.
Exemplary embodiments
The invention was made during the hd3 codec development. Therefore, the description is applicable to the current ver- sion of the hd3 bit stream format. However, the invention can be applied correspondingly to all frame based file for- mats that require knowledge of a number of previous frames for decoding a frame and wherein number is written into the bit stream header for initialising the decoder states, or for achieving sample-exact cutting by indicating the number of samples to mute of the decoded frame.
The hd3 codec processing enables storage, in a single file, of an mp3 file plus extension data for the bit-exact repre¬ sentation of the source of the mp3 file. The first version of the hd3 bit stream format was an archival format with one file header and successive data frames Frameø, Frame]_, ... Framen, as shown in Fig. 1. For an easier understanding, the different layers of the hd3 codec are not depicted here, and it is assumed that the described coder requires a number of previously decoded frames to decode the current frame.
For different applications it might be useful to transform the archival format at decoder side into a streaming format, or to cut a short sequence out of the archival format. For example, cutting enables fast generation of short previews and the streaming format allows starting decoding without having the complete file received.
However, a problem to solve for sample-exact cutting and for the operation in streaming mode and archival mode is the initialisation of the decoder states in case the previous frames are missing. The following description explains the required bit stream information and decoder processing.
Archival mode The archival mode bit stream as shown in Fig. 1 consists of one file header at the beginning of the file, followed by signal data divided into frames, wherein each frame repre¬ sents code representing a segment of L coded signal samples or coefficients. The archival mode bit stream decoding can start only at the first frame because only there the decoder states are known (they are defined in every decoding standard) . E.g., the be¬ low-mentioned Main_Data_Begin_Pointer in mp3 is set to zero for the first frame. By continuously decoding the following frames, the decoder states get initialised and a correct de- coding of the samples is obtained. A main characteristic of the described format is that not all but only k, with 0≤k<K, previously decoded frames are required to achieve correct results, wherein K is the maximal number of required decoded frames and K is a constant value that is known in the en- coder and in the decoder. However, decoding of the required information from the previous frames without knowledge of other frames needs to be feasible.
The hd3 codec provides such feature by the independently de- codable mp3 frames. However, the reconstruction of the bit- exact replica (i.e. lossless reconstruction) requires due to the mapping processing status information from three previously decoded mp3 frames, as described in application EP 08102308.7. Furthermore, the MPEGl Audio Layer III standard ISO/IEC
11172-3 includes the above-mentioned bit reservoir technique that allows storage of main data in previous frames. A main data begin pointer is used that points to the beginning of the main data located in a previous frame. Thus, decoding of single mp3 frames requires information from previous frames as well.
These issues are complicating the cutting or streaming of the said bit stream format due to the fact that the required information from previous frames is missing at the beginning of the streaming mode file or the cut file. Fig. 4 illus¬ trates related issues that affect decoding-side cutting or streaming of the encoding-side (or transmitted, or recorded, or original) archival format. When starting decoding at the header of the bit stream without the data from Framen_5, Framen_4 can not be decoded because its Main_Data_Begin_Pointer points to non-available Framen_5 data. As well, the frames n-3 to n-1 can not be decoded cor¬ rectly because their decoder states are not initialised cor¬ rectly due to the missing data from the previous frames. However, following availability of data (e.g. spectral val¬ ues) from three previous frames, although frames n-3 to n-1 are not decoded, the correct decoder states of the following frame can be reconstructed in a decoder states initialisa¬ tion step/stage DSI by performing the mapping process de¬ scribed in EP08102308.7. The first frame with correctly ini¬ tialised decoder states is Framen '
To overcome these problems, according to the invention, some extra information items are added to the bit stream header depicted in the following table. Based on these additional information items the decoder can decide if the processing will start with known default decoder states or with unknown decoder states:
Figure imgf000016_0001
Figure imgf000017_0001
Table 1: Section of the header of the inventive bit stream format
The xMuteIfFirstSuperFrame' field of the bit stream header indicates whether or not decoding of the following frames requires an additional initialisation of the decoder states, If true (i.e. the flag is set) , further information items are inserted into the bit stream header, which will be dis¬ cussed in the following examples for cutting and streaming.
The bit stream header for the archival mode uses the XOFI/ , 'FileMode' and 'MutelfFirstSuperFrame' data fields only, wherein the values xFileMode' and xMuteIfFirstSuperFrame' are set to zero (or the corresponding flags are cleared) to indicate a non-cut archival file.
Cutting of the archival mode bit stream
Cutting is used at decoding side to separate a short sub¬ section from a received or replayed complete coherent bit stream in archival mode. The bit stream mode of the cut file is equal to that of the archival mode format. Fig. 2 depicts a cut file produced from the archival mode file format shown in Fig. 1. Cutting can be used, for example, to get a short preview of the complete file. Instead of decoding and newly encoding the desired section, the required frames are just cut out from the input file and a new header is inserted in front of these cut-out data frames.
The problem with cutting is that for the decoding of the first frame of the cut-out data frames the data from the re¬ quired previous frames are missing in case the preview or the cut-out section, respectively, does not start at the be- ginning of the complete bit stream. Therefore the cut file evaluates the xMuteIfFirstSuperFrame' data field of the bit stream header indicating that the processing of the first frames merely initialises the decoder states and that the decoded samples of these frames are to be muted. The number of frames that are to be muted is indicated in the bit stream header in the xmuteFrames' field, and the number xmuteSamples' of samples that are to be muted is indicated in the header to enable sample-exact cutting instead of pure frame-exact cutting.
The following example explains in more detail how to create a cut file from an archival file:
Given values
L: Number of decoded samples per frame
OFLorig : Total length of the input file in samples per channel
Xstart • Number of the first sample of the archival file to be decoded for the cut file
Y Λend • Number of the last sample of the archival file to be decoded for the cut file
Values to be determined
OFLcut : Total length of the cut file in samples per chan- nel
Framevaild: First decodable frame with known decoder states
Framestart: First decoded frame of the cut file
Figure imgf000019_0001
Create the cut file opy the header followed by the frames Framestart to Frameend
For identification of a cut archival mode file, xFileMode' is set to x0' and 'MutelfFirstSuperFrame' is set to Λl' . These properties will indicate to the decoder that the de- coder states are to be initialised using the first
'muteFrame' frames parameter, and that the decoded samples of these frames are invalid.
Several frame-dependent values need to be found out for per- forming a cutting of the archival bit stream mode-file. How¬ ever, the first step is finding the frame xFramevaild' in which the first sample of the cut file is stored. For the computation of that first frame, corresponding delays in the core-codec processing may be introduced which, for easier understanding, are not considered in the example.
Next, the number of previous frames required for recovering the decoder states is to be obtained. Thus, the cut file must start with the frame that holds the data of the first frame involved in the decoder state recovery of xFramevaild' . The obtained number of previous frames is written into the 'muteFrame' field of the bit stream header, such that the decoder knows that these frames are to be used only for ini¬ tialisation of the decoder states and not for decoding samples . For enabling a sample-exact cutting, an additional number of 'samples per channel to mute' ( xmuteSamples' ) can be indi¬ cated in the header as well. These samples will be decoded correctly by the decoder but will not be presented to the user. Therefore the decoded presented signal can start with a sample located anywhere within the frame, instead of starting at the beginning of the frame.
In each case, XOFI/ is the number of samples per channel that can be decoded with known decoder states. This is ex¬ plained in more detail in the following Streaming bit stream mode section. Therefore the samples of the initialisation frames are to be added to the actual number of samples of the cut file. The decoder will automatically subtract these additional samples from the XOFL' value in the decoder proc¬ ess. Further details are described in the Decoding process¬ ing section.
Streaming mode
Using information items from the bit stream header depicted in Table 1, the archival bit stream mode can be transformed at decoding side into a streaming bit stream mode. 'Streaming mode' means that the frames of the archival bit stream are separated into successive packages, whereby each one of these packages is called a 'super frame' (SF) and has the same structure as an archival bit stream. A super frame starts with the corresponding bit stream section header
(i.e. a super frame header) followed by frames of data. I.e., in comparison with the archival mode, in the streaming mode super frames each one having a header are repeatedly arranged in the bitstream. An example for such a streaming mode bit stream is depicted in Fig. 3.
The xFileMode' data field of the bit stream header of each super frame carries the value xl' indicating the streaming mode, in which the decoder must decode several successive super frames in order to reconstruct all samples of the en¬ coded file. The first super frame of a streaming mode stream is basically identical to the beginning of an archival mode file, but in the bit stream header the data fields xFileMode' and XOFL' carry values different than in an ar¬ chival mode header. When the decoder starts decoding of the first super frame, the default decoder states are used and the first frame can be decoded directly without requiring any further information items. However, a streaming mode bit stream can be replayed or decoded starting at each one of its super frames. But in that case the required data from previous frames for initialising the decoder states are missing. Therefore all super frames of a stream except the first super frame must indicate in the 'MutelfFirstSuperFrame' data field that their first xmuteFrames' number of frames are to be used only for the decoder state recovery.
In streaming mode one must distinguish the decoding initialisation phase, where the decoding states are unknown, from the decoding phase, where the decoding states are known from previously decoded super frames. The corresponding type of phase is obtained at the beginning of the each super frame using the header information of the super frame. The header properties for both phases are shown in the following.
Decoding initialisation phase FileMode = 1
MutelfFirstSuperFrame = 0
AND decoding starts at the beginning of a new coherent bit stream in streaming mode.
The default decoder states are used and the decoding of the samples can start directly at the first frame of the super frame .
FileMode = 1 MutelfFirstSuperFrame = 1 AND this is the first super frame to decode following a de¬ coder initialisation or reset.
The first 'muteFrame' frames of the current super frame are used for initialising the decoder states, and the decoding of samples can start at frame number ('muteFrames' +1) .
FileMode = 1 MutelfFirstSuperFrame = 1
AND this is not the first decoded super frame AND the con¬ sistency check of the previous and current super frames has failed. For the consistency check the xStreamingCheckSum' data field of the header is used. Its value is obtained from the data of the previous super frame and is written into the 'Stream- ingCheckSum' data field of the following super frame, so that successive super frames can be identified. For example, a Cyclic Redundancy Check sum (CRC) or a hash value of frames, e.g. the last frame, of the previous super frame can be used. The decoder computes the CRC of the previous super frame and compares it with the value stored in the header of the current super frame. If this comparison fails the cur- rent decoder states are not valid for the decoding of the next super frame. Thus, the decoder states are to be re¬ initialised using the first xmuteFrames' number of frames of the next super frame. Decoding of samples starts at frame number ('muteFrames' +1) . The xStreamingCheckSum' data field is not used in the first super frame of a coherent streaming mode bit stream because previous frames are not required and therefore the consis¬ tency check is superfluous.
Decoding phase
FileMode = 1
MutelfFirstSuperFrame = 1
AND a super frame was decoded before AND the consistency check is valid. Thus, the decoder status of the previously decoded super frame can be used to decode the frames of the next super frame .
A typical streaming processing starts with an initialisation phase followed by several decoding phases. In the decoding phase the 'muteFrame' information is not used because the decoder states are initialised correctly by corresponding data from the previous super frame, and the data to which the xMain_Data_Begin_Pointer' is pointing is available in the previous super frame. The decoder will only return to the initialisation phase if a new coherent bit stream will start (MutelfFirstSuperFrame =0) or the xStreamingCheckSum' is incorrect. In both cases the decoder states become inva¬ lid and are to be re-initialised.
The following example shows how to create a streaming mode bit stream from an already existing archival mode bit stream:
Given values
L: Number of decoded samples per frame
OFLor-j_g: Total length of the input file in samples per chan¬ nel M: Number of super frames with M ≤ ceil (OFLor-j_g/L)
Values to be determined
NumFrames: Total number of frames
Mmean: Average number of frames per super frame
Frames-|-ar-|- : First frame of a super frame
Frameen(j: Last frame of a super frame muteFramesgp : Number of frames for initialisation of the decoder states of the super frame StreamingCheckSum5p : Value for consistency check of the super frame m: Number of currently written super frames with 0 < m < M
Calculate total number of frames
NumFrames = ceil (OFLor-j_g/L) , wherein xceil' returns a value representing the smallest integer that is greater equal the argument of the ceil function.
Calculate number of frames per super frame Mmean = ceil (0FLorig/ (M*L) ) Create first super frame (m=0)
Establish the corresponding header:
Figure imgf000025_0001
Copy frames number x0' to (Mmean-1) .
Create following super frames (l≤m<M)
Compute xStreamingCheckSum' : compute the Cyclic Redundancy Check sum of a defined part of the previous super frame and store the result in data field ^StreamingCheckSuir^p' .
Obtain xmuteFrames' : find all frames with an invalid Main_Data_Begin_Pointer that points to data not located in the current super frame and add the number of frames for re¬ covering the decoder states. Set 'muteFramesgp' to the result of this sum.
Establish the corresponding header:
Figure imgf000025_0002
Copy frames number (m*Mmean) to min ( ( (m+1) *Mmean) -1, NumFrames -1), wherein the min function is required for the last super frame because the last super frame could carry a different number of frames and samples. Decoding processing for the file or bit stream The decoding processing flow chart of Fig. 5 describes the decoding process for archival mode files and for streaming mode bit streams. At the beginning of the decoding process the decoder states are set to their default values (e.g. a mapping buffer is set to zero and previously decoded values are set to zero) . In step/stage 1 the input file or bit stream is available for reading the required data. Step/stage 2 finds and reads the first header of the file or bit stream, i.e. the header information items of Table 1 are set, stored or loaded. Step/stage 3 sets the variable xSam- plesToMute' to its default value zero.
In step/stage 4 the xMuteIfFirstSuperFrame' data field of the file or bit stream header is checked for deciding whether the decoder states are known (MutelfFirstSuperFrame =0), or the decoder states must be re-initialised and a num¬ ber of samples are to be ignored. The number of ignored (muted) samples is calculated in step/stage 5 as Samples- ToMute = muteFrames*L + muteSamples from the number xmuteFrames' of frames to mute and the number xmuteSamples' of samples to mute received from the file or bit stream header. XI/ is the number of samples that are decoded from one frame, which is a known decoder constant. Step/stage 6 decodes all received frames which includes also the first xmuteFrames' frames. Hence XOFI/ decoded samples are passed to the next step/stage. Although the first xSamplesToMute' samples are invalid, they are used for initialising the de¬ coder states. Hence the following step/stage 7 removes the invalid xSam- plesToMute' samples and returns only the remaining samples to step/stage 8. The number of remaining samples is the dif¬ ference between the value xSamplesToMute' and the total num¬ ber OFL of decoded samples. In case that xSamplesToMute' is greater than OFL, zero samples are returned to step/stage 8. This can happen in the streaming-mode if the number of frames per super frame is less than the number of frames to mute .
Therefore the remaining delay is to be transferred to the next super frame. A corresponding variable MuteNextSF = SamplesToMute - OFL is calculated in step/stage 9 in order to store the number of samples to mute in the next super frame .
At step/stage 10 the decoding processing is finished 11 for an archival mode file or bit stream or a cut file bit stream because all OFL samples of the file have been decoded.
Thus, step/stage 10 checks the xFileMode' and proceeds to END step 11 for stopping the archival mode decoding process. Otherwise, decoding of the streaming-mode bit stream is to be continued. Therefore the xStreamingCheckSum' , for example a CRC, is computed from the frame data of the currently decoded frame in step/stage 12. The processing and the data used for the xStreamingCheckSum' computation must produce identical results in the encoder and the decoder. Furthermore, the xStreamingCheckSum' should represent a clear identification of the current frame because it is used for verifying the consistency of the decoder states of the next super frame. Therefore the used data should diversify from super frame to super frame and represent the coded data of the super frame.
Upon switching to the next super frame, step/stage 13 searches and reads the header of the next super frame. This step/stage re-initialises all bit stream header variables so that the former header information is lost. Upon reaching the end of the file, or in case no valid bit stream header has been found, the decoder will proceed to step 11 and stop the decoding.
Otherwise it is to be checked for decoding the next super frame whether the current decoder states are valid. A streaming mode bit stream file can consists of two or more successive coherent bit streams. Therefore it is to be checked whether the following super frame is the first super frame of a new coherent bit stream, because in such case the current decoder states belong to a different bit stream and are to be reset to the default values. The first super frame of a coherent bit stream has a xMuteIfFirstSuperFrame' value of zero. Therefore the xMuteIfFirstSuperFrame' value is in¬ verted in step/stage 14 and is assigned to the variable xFirstSuperFrame' , i.e. FirstSuperFrame = NOT (MutelfFirstSuperFrame) . Step/stage 15 checks variable xFirstSuperFrame' whether a new coherent bit stream is starting. If true, step/stage 17 resets the decoder states and starts decoding of the next super frame at step/stage 3. This is identical to starting decoding of a new file, and is also identical to the decoder initialisation phase 1 of the streaming-mode section in this description. Otherwise the next super frame belongs to the same coherent bit stream and the processing continues in step/stage 16. Here, the xStreamingCheckSum' value of the current super frame bit stream header is compared with the xStreaming-
CheckSum' value computed from the previous super frame data in step/stage 12, i.e. is CRC == StreamingCRC ? This is necessary because a super frame of the current co¬ herent bit stream might be missing, or a new coherent bit stream is starting but not at its first super frame. In case the consistency-check of step/stage 16 fails, step/stage 17 is used to reset the decoder states and then decoding of the next super frame starts at step/stage 3. Therefore the processing is in the decoder initialisation phase 3 of the streaming-mode section in this description and the decoding is started with default decoder states, which leads to the decoder initialisation phase 2 in step/stage 4. Otherwise the current decoder states are valid for decoding of the current super frame and the current processing is in the decoding phase 1 of the streaming-mode section in this description. Consequently, the 'muteFrames' value of the current bit stream header needs not be evaluated. This is ensured by step/stage 18 which sets the xMuteIfFirstSuper- Frame' variable to zero. Therefore step/stage 4 will not lead to step/stage 8 and the xmuteFrames' value of the cur¬ rent super frame will not be used.
The decision of step/stage 19 is made for transferring a remaining delay of the previous super frame to the current super frame. If ^MuteNextSF' > 0, the number of samples to mute of the previous super frame was greater than the XOFI/ value of the previous super frame. Consequently the number xSamplesToMute' of samples to mute of the current super frame is equal to the remaining number ^MuteNextSF' of samples to mute of the previous super frame. Therefore step/stage 20 sets SamplesToMute = MuteNextSF .
Accordingly, decoding of the current super frame must start at step/stage 4 in order to omit the re-initialisation of xSamplesToMute' in step/stage 3. If in step/stage 19 parameter MuteNextSF < 0, there are no remaining samples to mute and the decoder states are correct and the decoding of the current super frame can start di¬ rectly with the first frame. The number of samples to mute is set to zero in step/stage 3. In Streaming mode the decoding of successive super frames is repeated until step/stage 13 does no more find a bit stream header, and the decoding processing is stopped in step 11.
Thus, the invention facilitates processing of a frame based bit stream format in archival mode bit stream or streaming mode bit stream, and enables a sample-exact cutting of the archival mode bit stream, even if decoding of one frame re¬ quires information from previous frames. As mentioned above, the inventive decoding processing is used in a hd3 decoder implementation and is tested successfully.
A HD3 file in it simplest format has three mandatory data sections shown in Fig. 6, and up to four optional data sec¬ tions. The mandatory hd3lD data section provides file-mode, length, and CRC information. This hd3lD section is encapsulated in an ancillary data section (silence PCM) of an mp3 frame. Such ancillary data section of an mp3 frame may also contain a Xing- or VBRi-compliant variable bit rate header in front of the hd3lD information.
The mandatory mp3-data section encapsulates the mp3 data stream or parts thereof. The synchronisation words of the mp3 stream may be scrambled.
The mandatory cd-data section provides the lossless exten¬ sion to the mp3 data for reconstructing a mathematically lossless copy of the original music with one or two audio channels having a temporal resolution corresponding to 32 kHz, 44.1 kHz or 48 kHz and an amplitude resolution of 16 bit/sample. An unsynchronisation scheme can be used that avoids the occurrence of mp3 sync headers.
An optional hd-data section (not depicted) enables lossless reconstruction of high-definition studio formats with up to 24 bit/sample and sampling rates of 192 kHz.
The optional ID3/ID3v2 data section stores metadata relating to the embedded mp3 encoded and the lossless reconstruction encoded music. There may also be an optional ID3vl tag at the very end of the HD3 file (not depicted) .
The invention can also be used for other codecs or bit streams like AAC, and in mp3 where the additional data (e.g. the headers) are arranged in mp3 ancillary data fields, which a corresponding mp3 decoder can evaluate.

Claims

Claims
1. Method for generating a frame based bit stream format file including at least one bit stream header section (Header) , said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame (Frameg, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), characterised by arranging in said header information items representing: total number of samples per channel (OFL) for said file; an 'archival mode' or a 'streaming mode' (FileMode) for said file; - whether or not a decoding of said encoded signal data can start at the first frame of said file (MutelfFirstSuperFrame) ; if a decoding of said encoded signal data is not to start at said first frame of said file, the number of initial frames to mute (muteFrames) while decoding states are initialised, and optionally, for a sample- exact cutting feature, the number of samples to mute (muteSamples) in the corresponding frame; in said streaming mode, a value (StreamingCheckSum) for identifying a previous super frame for a consistency check, wherein in said archival mode, said file includes a sin¬ gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , and a decoding of said encoded signal data is to start with the first one (Frameg) of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section (Header) followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute (muteFrames) signalises how many ones of initial frames in a current super frame are re- quired for establishing data for initialising states of a decoding of said signal data before actually starting de¬ coding of encoded signal data from the following frame or frames of said current super frame, such initial frames not being used for decoding the encoded signal data con- tained therein.
2. Data structure for a frame based bit stream format file including at least one bit stream header section (Header) , said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame (Frameg, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), characterised by having in said header information items representing: - total number of samples per channel (OFL) for said file; an 'archival mode' or a 'streaming mode' (FileMode) for said file; whether or not a decoding of said encoded signal data can start at the first frame of said file (MutelfFirstSuperFrame) ; if a decoding of said encoded signal data is not to start at said first frame of said file, the number of initial frames to mute (muteFrames) while decoding states are initialised, and optionally, for a sample- exact cutting feature, the number of samples to mute (muteSamples) in the corresponding frame; in said streaming mode, a value (StreamingCheckSum) for identifying a previous super frame for a consistency check, wherein in said archival mode, said file includes a sin- gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , and a decoding of said encoded signal data is to start with the first one (Frameg) of said frames, and wherein in said streaming mode said bit stream contains more than one super frame, each one of said super frames starting with a single bit stream header section (Header) followed by several ones of said frames, and a decoding of said encoded signal data can be initialised at each one of said super frames, and wherein said information item regarding the number of initial frames to mute (muteFrames) signalises how many ones of initial frames in a current super frame are re¬ quired for establishing data for initialising states of a decoding of said signal data before actually starting de¬ coding of encoded signal data from the following frame or frames of said current super frame, such initial frames not being for use for decoding the encoded signal data contained therein.
3. Method for cutting a section out of a frame based bit stream format file including a bit stream header section (Header) , said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame (Frameø, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), and wherein said header includes information items represent¬ ing: total number of samples per channel (OFL) for said file; an 'archival mode' (FileMode) for said file; a decoding of said encoded signal data is to start at the first frame of said file (MutelfFirstSuperFrame) , wherein in said archival mode, said file includes a sin- gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , said method including the step of forming a cutting file by taking the required frames data from said bit stream and arranging a cutting header in front of these frames data, wherein said cutting header is derived from said header and in said cutting header: said total number of samples per channel (OFL) for said file is replaced by the number of samples per channel (OFLCU-|-) for said cutting file; said archival mode (FileMode) is kept; - said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is changed to decoding of said encoded signal data is to start later in said file; an information item is added regarding the number of ini- tial frames to mute (muteFrames) while decoding states are initialised.
4. Apparatus for cutting a section out of a frame based bit stream format file including a bit stream header section (Header) , said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame (Frameg, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), and wherein said header includes information items represent- ing: total number of samples per channel (OFL) for said file; an 'archival mode' (FileMode) for said file; a decoding of said encoded signal data is to start at the first frame of said file (MutelfFirstSuperFrame) , wherein in said archival mode, said file includes a sin¬ gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , said apparatus including means being adapted for forming a cutting file by taking the required frames data from said bit stream and arranging a cutting header in front of these frames data, wherein said cutting header is de¬ rived from said header, and in said cutting header: said total number of samples per channel (OFL) for said file is replaced by the number of samples per channel (OFLCU-|-) for said cutting file; said archival mode (FileMode) is kept; said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is changed to decoding of said encoded signal data is to start later in said file; an information item is added regarding the number of initial frames to mute (muteFrames) while decoding states are initialised.
5. Method according to claim 3, or apparatus according to claim 4, wherein said cutting header additionally includes an information item (muteSamples) regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin.
6. Method according to claim 3 or 5, or apparatus according to claim 4 or 5, wherein said information items in said cutting header are calculated as follows:
Frameva]_j_(j = floor (Xstart /L) ' wherein Xstart i-s ^he num- ber of the first encoded signal sample or coefficient of the archival file to be decoded for the cut file and L is the number of samples or coefficients per frame;
Get the number kmapp-j_ng of frames required for mapping status initialisation and a Main_Data_Begin_Pointer of the first decodable frame Frameva]_-j_^ with known decoder states;
Set the number kmp3 of invalid frames equal to the number of frames to add in order to get a valid
Main_Data_Begin_Pointer; - said information item regarding the number of frames to mute is muteFrames = kmp3 + kmapp-j_nq; said number of samples per channel for said cutting file is OFLCU-|- = Xend ~ xstart +1 + L*muteFrames, wherein Xend is the number of the last encoded signal sample or coef¬ ficient of the archival file to be decoded for the cut file.
7. Method for changing a frame based bit stream file format, said bit stream including a header section (Header) , said frames including encoded signal data, wherein data re- quired for a decoding or an evaluation of data of a current frame (Frameg, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), and wherein said header includes information items representing: total number of samples per channel (OFL) for said file; an 'archival mode' (FileMode) for said file; a decoding of said encoded signal data is to start at the first frame of said file (MutelfFirstSuperFrame) , wherein in said archival mode, said file includes a sin- gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , said method including the step of generating from said archival mode file a 'streaming mode' bit stream by form¬ ing from every successive group of frames of said archi- val mode file a super frame of said streaming mode bit stream by taking the required frame data from said archi¬ val mode file and arranging a super frame header at the beginning of each super frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers: said total number of samples per channel (OFL) for said file is replaced by a first adapted number of samples per channel; an information item for 'streaming mode' instead of 'archival mode' is given; said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is kept, and in the following ones of said super frame headers: said total number of samples per channel (OFL) for said file is replaced by a second adapted number of samples per channel; said information item for 'archival mode' (FileMode) is replaced by an information item for 'streaming mode' ; said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is replaced by an information item regarding decoding of said encoded signal data is not to start at said first frame; an information item is added regarding the number of initial frames to mute (muteFrames) while decoding states are initialised; optionally, an information item (muteSamples) is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is O'; an information item (StreamingCheckSum) is added regarding consistency check data, the value of which is derived from the data of the previous super frame.
8. Apparatus for changing a frame based bit stream file for¬ mat, said bit stream including a header section (Header) , said frames including encoded signal data, wherein data required for a decoding or an evaluation of data of a current frame (Frameg, Framen) can be contained in one or more previous frames (Framen_5, Framen_4), and wherein said header includes information items representing: total number of samples per channel (OFL) for said file; an 'archival mode' (FileMode) for said file; - a decoding of said encoded signal data is to start at the first frame of said file (MutelfFirstSuperFrame) , wherein in said archival mode, said file includes a sin¬ gle bit stream header section (Header) and successive ones of said frames (Frameg, Frame]_, ..., Framen) , said apparatus including means being adapted for generat- ing from said archival mode file a 'streaming mode' bit stream by forming from every successive group of frames of said archival mode file a super frame of said stream¬ ing mode bit stream by taking the required frame data from said archival mode file and arranging a super frame header at the beginning of each super frame, wherein these super frame headers are derived from said header and in the first one of said super frame headers: said total number of samples per channel (OFL) for said file is replaced by a first adapted number of samples per channel; an information item for 'streaming mode' instead of 'archival mode' is given; said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is kept, and in the following ones of said super frame headers: said total number of samples per channel (OFL) for said file is replaced by a second adapted number of samples per channel; - said information item for 'archival mode' (FileMode) is replaced by an information item for 'streaming mode' ; said information item (MutelfFirstSuperFrame) regarding decoding of said encoded signal data is to start at the first frame of said file is replaced by an information item regarding decoding of said encoded signal data is not to start at said first frame; an information item is added regarding the number of initial frames to mute (muteFrames) while decoding states are initialised; - optionally, an information item (muteSamples) is added regarding the number of samples to mute in the frame where decoding of said encoded signal data is to begin, which number is x0'; an information item (StreamingCheckSum) is added regarding consistency check data, the value of which is derived from the data of the previous super frame.
9. Method according to claim 7, or apparatus according to claim 8, wherein said first adapted number of samples per channel is OFL = Mmean*L, wherein L is the number of de- coded samples per frame, OFLor-j_g is the total length of said file in samples per channel, M is the number of su¬ per frames with M ≤ ceil (OFLor-j_g/L) , and Mmean = ceil (0FLorig/ (M*L) ) .
10. Method or apparatus according to claim 9, wherein xm' is the number of a current super frame and said second adapted number of samples per channel is OFL = min ( (Mmean*L) , (0FLorig - m*L*Mmean) ) .
11. Method or apparatus according to claim 10, wherein said information item (StreamingCheckSum) regarding consistency check data is the Cyclic Redundancy Check sum of a pre-defined section of the previous super frame.
12. Method or apparatus according to claim 10, wherein said information item (StreamingCheckSum) regarding consistency check data is the hash value of frames, e.g. the last frame, of the previous super frame.
13. Use of the method according to one of claims 7, 9 and 10 for decoding said streaming mode bit stream, further including the steps of: when starting decoding of said streaming mode bit stream at said first super frame, the decoding of the encoded signal data starts at the first frame of that super frame using default decoder states; when not starting decoding of said streaming mode bit stream at said first super frame, following a decoding initialisation or reset, a number xmuteFrames' of frames corresponding to said information item regarding the number of initial frames to mute (muteFrames) is used for initialising the decoding states, and the decoding of encoded signal data starts at frame number muteFrames+1 of the current super frame using these de¬ coding states; - when not starting decoding of said streaming mode bit stream at said first super frame, and consistency check data that were calculated from the previous super frame data is not consistent with corresponding consistency check data calculated from the current super frame, us- ing a number 'muteFrames' of frames of the following su¬ per frame for re-initialising the decoding states starting decoding of encoded signal data at frame number 'muteFrames' +1 of that following super frame using these decoding states; - when not starting decoding of said streaming mode bit stream at said first super frame, and a super frame was decoded before and said consistency check data are valid, using the decoder states of the previously decoded super frame to decode the encoded signal data of the frames of the current super frame.
PCT/EP2009/055663 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure WO2009146998A1 (en)

Priority Applications (7)

Application Number Priority Date Filing Date Title
JP2011510935A JP5774983B2 (en) 2008-06-02 2009-05-11 Method and apparatus for generating, cutting or modifying a frame-based bitstream format file including at least one header section and corresponding data structure
US12/737,007 US8731946B2 (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
EP09757381A EP2289245B1 (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
PL09757381T PL2289245T3 (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
CN200980120515.2A CN102047336B (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
ES09757381T ES2379321T3 (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame-based bit stream format file that includes at least one header section, and a corresponding data structure
AT09757381T ATE545281T1 (en) 2008-06-02 2009-05-11 METHOD AND DEVICE FOR PRODUCING OR FOR CUTTING OR MODIFYING A FRAME BASED FILE IN BIT STREAM FORMAT WITH AT LEAST ONE HEAD PART AND CORRESPONDING DATA STRUCTURE

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP08157417A EP2131590A1 (en) 2008-06-02 2008-06-02 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
EP08157417.0 2008-06-02

Publications (1)

Publication Number Publication Date
WO2009146998A1 true WO2009146998A1 (en) 2009-12-10

Family

ID=40193935

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2009/055663 WO2009146998A1 (en) 2008-06-02 2009-05-11 Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure

Country Status (10)

Country Link
US (1) US8731946B2 (en)
EP (2) EP2131590A1 (en)
JP (1) JP5774983B2 (en)
KR (1) KR20110026445A (en)
CN (1) CN102047336B (en)
AT (1) ATE545281T1 (en)
ES (1) ES2379321T3 (en)
PL (1) PL2289245T3 (en)
TW (1) TWI469134B (en)
WO (1) WO2009146998A1 (en)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1881485A1 (en) * 2006-07-18 2008-01-23 Deutsche Thomson-Brandt Gmbh Audio bitstream data structure arrangement of a lossy encoded signal together with lossless encoded extension data for said signal
EP2395505A1 (en) 2010-06-11 2011-12-14 Thomson Licensing Method and apparatus for searching in a layered hierarchical bit stream followed by replay, said bit stream including a base layer and at least one enhancement layer
JP5749462B2 (en) * 2010-08-13 2015-07-15 株式会社Nttドコモ Audio decoding apparatus, audio decoding method, audio decoding program, audio encoding apparatus, audio encoding method, and audio encoding program
KR20120074059A (en) * 2010-12-27 2012-07-05 삼성전자주식회사 Method and apparatus for outputting audio data
CN102522088B (en) * 2011-11-25 2013-10-23 展讯通信(上海)有限公司 Decoding method and device of audio frequency
EP2863386A1 (en) * 2013-10-18 2015-04-22 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Audio decoder, apparatus for generating encoded audio output data and methods permitting initializing a decoder
CA2973464C (en) * 2015-01-20 2019-11-19 Sharp Kabushiki Kaisha Service guide encapsulation
EP3438976A4 (en) * 2016-03-31 2019-04-24 Sony Corporation Information processing device and method
US10135899B1 (en) * 2016-12-16 2018-11-20 Amazon Technologies, Inc. Dynamic archiving of streaming content
CN109584891B (en) * 2019-01-29 2023-04-25 乐鑫信息科技(上海)股份有限公司 Audio decoding method, device, equipment and medium in embedded environment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020097272A1 (en) * 2001-01-24 2002-07-25 Mitsumasa Tanaka Image editing apparatus and image editing method
WO2003038814A2 (en) * 2001-10-31 2003-05-08 Thomson, Licensing. S. A. Changing a playback speed for video presentation recorded in a modified film format
EP1830575A2 (en) * 2006-03-01 2007-09-05 Sony Corporation Compressed video data processing apparatus and method
US20070280438A1 (en) * 2006-05-17 2007-12-06 Recording For The Blind & Dyslexic, Incorporated Method and apparatus for converting a daisy format file into a digital streaming media file
US20080125891A1 (en) * 2006-11-29 2008-05-29 Matsushita Electric Industrial Co., Ltd. Audio reproduction apparatus, audio reproduction method, program, and data management method

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5677918A (en) * 1995-07-28 1997-10-14 Motorola, Inc. Method and device for efficient error correction in a packet-switched communication system
GB0007870D0 (en) * 2000-03-31 2000-05-17 Koninkl Philips Electronics Nv Methods and apparatus for making and replauing digital video recordings, and recordings made by such methods
EP1215663A1 (en) * 2000-12-15 2002-06-19 BRITISH TELECOMMUNICATIONS public limited company Encoding audio signals
JP3871210B2 (en) * 2002-09-19 2007-01-24 ソニー株式会社 CONVERTING APPARATUS, CONVERTING METHOD, PROGRAM, AND DATA STRUCTURE
EP1647010B1 (en) * 2003-07-21 2017-09-06 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Audio file format conversion
US8233462B2 (en) * 2003-10-15 2012-07-31 Qualcomm Incorporated High speed media access control and direct link protocol
US8131134B2 (en) * 2004-04-14 2012-03-06 Microsoft Corporation Digital media universal elementary stream
KR100846781B1 (en) * 2004-06-01 2008-10-27 삼성전자주식회사 Service display control method using a Fast Information Channel in DAB receiver, and apparatus thereof
US7672743B2 (en) * 2005-04-25 2010-03-02 Microsoft Corporation Digital audio processing
JP2006317575A (en) * 2005-05-11 2006-11-24 Matsushita Electric Ind Co Ltd Audio decoding device
KR100754197B1 (en) * 2005-12-10 2007-09-03 삼성전자주식회사 Video service providing and receiving method in DAB system, and apparatus thereof
EP1855436A1 (en) * 2006-05-12 2007-11-14 Deutsche Thomson-Brandt Gmbh Method and apparatus for encrypting encoded audio signal
JP2008185769A (en) * 2007-01-30 2008-08-14 Oki Electric Ind Co Ltd Compressed audio reproduction device
US7778839B2 (en) * 2007-04-27 2010-08-17 Sony Ericsson Mobile Communications Ab Method and apparatus for processing encoded audio data

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020097272A1 (en) * 2001-01-24 2002-07-25 Mitsumasa Tanaka Image editing apparatus and image editing method
WO2003038814A2 (en) * 2001-10-31 2003-05-08 Thomson, Licensing. S. A. Changing a playback speed for video presentation recorded in a modified film format
EP1830575A2 (en) * 2006-03-01 2007-09-05 Sony Corporation Compressed video data processing apparatus and method
US20070280438A1 (en) * 2006-05-17 2007-12-06 Recording For The Blind & Dyslexic, Incorporated Method and apparatus for converting a daisy format file into a digital streaming media file
US20080125891A1 (en) * 2006-11-29 2008-05-29 Matsushita Electric Industrial Co., Ltd. Audio reproduction apparatus, audio reproduction method, program, and data management method

Also Published As

Publication number Publication date
TW200951941A (en) 2009-12-16
EP2289245A1 (en) 2011-03-02
ATE545281T1 (en) 2012-02-15
US8731946B2 (en) 2014-05-20
EP2289245B1 (en) 2012-02-08
KR20110026445A (en) 2011-03-15
CN102047336B (en) 2014-05-07
PL2289245T3 (en) 2012-05-31
US20110158326A1 (en) 2011-06-30
JP2011523090A (en) 2011-08-04
CN102047336A (en) 2011-05-04
EP2131590A1 (en) 2009-12-09
JP5774983B2 (en) 2015-09-09
TWI469134B (en) 2015-01-11
ES2379321T3 (en) 2012-04-24

Similar Documents

Publication Publication Date Title
WO2009146998A1 (en) Method and apparatus for generating or cutting or changing a frame based bit stream format file including at least one header section, and a corresponding data structure
EP1444689B1 (en) Determination of the presence of additional coded data in a data frame
US7974837B2 (en) Audio encoding apparatus, audio decoding apparatus, and audio encoded information transmitting apparatus
JP5249214B2 (en) Bitstream data of lossy encoded signal and audio bitstream data structure arrangement of lossless extended encoded data of the above signal
US8428941B2 (en) Method and apparatus for lossless encoding of a source signal using a lossy encoded data stream and a lossless extension data stream
JP2004046179A (en) Audio decoding method and device for decoding high frequency component by small calculation quantity
US20080288263A1 (en) Method and Apparatus for Encoding/Decoding
KR100904626B1 (en) Trick play for MP3
JP4359499B2 (en) Editing audio signals
JP2002368722A (en) Encoder and decoder
KR20080010980A (en) Method and apparatus for encoding/decoding
JP2009134115A (en) Decoder
JP2003233396A (en) Decoding device
JP2001175285A (en) Encoded sound reproduction device and encoded sound reproduction method

Legal Events

Date Code Title Description
WWE Wipo information: entry into national phase

Ref document number: 200980120515.2

Country of ref document: CN

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 09757381

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2009757381

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 2011510935

Country of ref document: JP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 20107029770

Country of ref document: KR

Kind code of ref document: A