Audio entropy encoder/decoder with different spectral resolutions and transform lengths and upsampling and/or downsampling
An audio encoder for encoding segments of coefficients, the segments of coefficients representing different time or frequency resolutions of a sampled audio signal, the audio encoder including a processor for deriving a coding context for a currently encoded coefficient of a current segment based on a previously encoded coefficient of a previous segment, the previously encoded coefficient representing a different time or frequency resolution than the currently encoded coefficient. The audio encoder further includes an entropy encoder for entropy encoding the current coefficient based on the coding context to obtain an encoded audio stream.
1. An audio encoder comprising a processor and a computer program stored on a digital storage medium configured to:
spectrally decompose, in a segment-wise manner and using different transform lengths, a sampled audio signal to obtain a sequence of segments of coefficients such that segments of coefficients for which different transform lengths are used, spectrally represent the sampled audio signal at different spectral resolutions and comprise different numbers of coefficients;
entropy encode a currently encoded coefficient of a current segment having a first spectral resolution based on an entropy coding context derived from a previously encoded coefficient of a previous segment having a second spectral resolution which is different from the first spectral resolution,
wherein the current segment is spectrally subdivided at the first spectral resolution into a first sequence of tuples of a first number spectrally neighbouring coefficients,
wherein the previous segment is spectrally subdivided at the second spectral resolution into a second sequence of tuples of a second number of spectrally neighbouring coefficients, the first number being equal to the second;
compute the entropy coding context for a current tuple comprising the currently encoded coefficient by:
downsampling the second sequence of tuples of the previous segment when the second spectral resolution is finer than the first spectral resolution, and/or up-sampling the second sequence of tuples of the previous segment, when the second spectral resolution is coarser than the first spectral resolution so as to obtain a third sequence of tuples so that
a tuple of the third sequence at an index k equals a tuple of the second sequence at an index k*ratio, wherein ratio corresponds to the second resolution divided by the first resolution,
selecting a set of tuples out of the third sequence, and
computing the entropy coding context for the current tuple on the basis of the set of tuples; and
entropy encode the currently encoded coefficient by entropy encoding the current tuple using the entropy coding context for the current tuple.
2. Method for encoding, comprising the steps of:
spectrally decomposing, in a segment-wise manner and using different transform lengths, a sampled audio signal to obtain a sequence of segments of coefficients such that segments of coefficients for which different transform lengths are used, spectrally represent the sampled audio signal at different spectral resolutions and comprise different numbers of coefficients;
entropy encoding a currently encoded coefficient of a current segment having a first spectral resolution based on an entropy coding context derived from a previously encoded coefficient of a previous segment having a second spectral resolution which is different from the first spectral resolution,
wherein the current segment is spectrally subdivided at the first spectral resolution into a first sequence of tuples of a first number spectrally neighbouring coefficients,
wherein the previous segment is spectrally subdivided at the second spectral resolution into a second sequence of tuples of a second number of spectrally neighbouring coefficients, the first number being equal to the second number;
computing the entropy coding context for a current tuple comprising the currently encoded coefficient by:
downsampling the second sequence of tuples of the previous segment when the second spectral resolution is finer than the first spectral resolution, and/or up-sampling the second sequence of tuples of the previous segment, when the second spectral resolution is coarser than the first spectral resolution so as to obtain a third sequence of tuples so that
a tuple of the third sequence at an index k equals a tuple of the second sequence at an index k*ratio, wherein ratio corresponds to the second resolution divided by the first resolution,
selecting a set of tuples out of the third sequence, and
computing the entropy coding context for the current tuple on the basis of the set of tuples; and
entropy encoding the currently encoded coefficient by entropy encoding the current tuple using the entropy coding context for the current tuple.
3. An audio decoder comprising a processor and a computer program stored on a digital storage medium configured to:
receive an encoded audio stream into which an audio signal is coded in a manner spectrally decomposed, in a segment-wise manner and using different transform lengths, thereby obtaining a sequence of segments of coefficients, wherein segments of coefficients for which different transform lengths are used, spectrally represent the audio signal at different spectral resolutions and comprise different numbers of coefficients;
entropy decode a currently decoded coefficient of a current segment having a first spectral resolution based on an entropy coding context derived from a previously decoded coefficient of a previous segment having a second spectral resolution which is different from the first spectral resolution,
wherein the current segment is spectrally subdivided at the first spectral resolution into a first sequence of tuples of a first number of spectrally neighbouring coefficients,
wherein the previous segment is spectrally subdivided at the second spectral resolution into a second sequence of tuples of a second number of spectrally neighbouring coefficients, the first number being equal to the second number;
compute the entropy coding context for a current tuple comprising the currently decoded coefficient by:
downsampling the second sequence of tuples of the previous segment when the second spectral resolution is finer than the first spectral resolution, and/or up-sampling the second sequence of tuples of the previous segment, when the second spectral resolution is coarser than the first spectral resolution so as to obtain a third sequence of tuples so that
a tuple of the third sequence at an index k equals a tuple of the second sequence at an index k*ratio, wherein ratio corresponds to the second resolution divided by the first resolution,
selecting a set of tuples out of the third sequence, and
computing the entropy coding context for the current tuple on the basis of the set of tuples; and
entropy decode the currently decoded coefficient by entropy decoding the current tuple using the entropy coding context for the current tuple.
4. An audio decoding method, comprising:
receiving an encoded audio stream into which an audio signal is coded in a manner spectrally decomposed, in a segment-wise manner and using different transform lengths, thereby obtaining a sequence of segments of coefficients, wherein segments of coefficients for which different transform lengths are used, spectrally represent the audio signal at different spectral resolutions and comprise different numbers of coefficients;
entropy decoding a currently decoded coefficient of a current segment having a first spectral resolution based on an entropy coding context derived from a previously decoded coefficient of a previous segment having a second spectral resolution which is different from the first spectral resolution,
wherein the current segment is spectrally subdivided at the first spectral resolution into a first sequence of tuples of a first number spectrally neighbouring coefficients,
wherein the previous segment is spectrally subdivided at the second spectral resolution into a second sequence of tuples of a second number spectrally neighbouring coefficients, the first number being equal to the second number;
computing the entropy coding context for a current tuple comprising the currently decoded coefficient by:
downsampling the second sequence of tuples of the previous segment when the second spectral resolution is finer than the first spectral resolution, and/or up-sampling the second sequence of tuples of the previous segment, when the second spectral resolution is coarser than the first spectral resolution so as to obtain a third sequence of tuples so that
a tuple of the third sequence at an index k equals a tuple of the second sequence at an index k*ratio, wherein ratio corresponds to the second resolution divided by the first resolution,
selecting a set of tuples out of the third sequence, and
computing the entropy coding context for the current tuple on the basis of the set of tuples; and
entropy decoding the currently decoded coefficient by entropy decoding the current tuple using the entropy coding context for the current tuple.
5. A computer program product stored in a non-transitory computer-readable medium, said computer program product having a program code for performing one of the methods of claim 2 or 4 , when the program code runs on a computer or a processor.