Multi-channel predictive subband audio coder using psychoacoustic adaptive bit allocation in frequency, time and over the multiple channels

A multi-channel audio encoder, comprising:

• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames, said multi-channel audio signal being encoded at a known bit rate;
• a plurality of filters comprising non-perfect and perfect reconstruction filters that are used to split the audio frames into respective pluralities of frequency subbands over a baseband frequency range when the known bit rate is respectively below and above a threshold bit rate, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a plurality of subband encoders that code the audio data in the respective frequency subbands a subframe at a time to produce encoded subband signals;
• an analyzer that splits each subframe in the audio window into a plurality of sub-subframes, detects transients in each sub-subframe, generates a transient code that indicates whether there is a transient in any sub-subframe other than the first and in which sub-subframe the transient occurs, and when a transient is detected generates a pre-transient scale factor for those sub-subframes before the transient and a post-transient scale factor for those sub-subframes including and after the transient and otherwise generates a uniform scale factor for the subframe,
• said subband encoders using said pre-transient, post-transient and uniform scale factors to scale the audio data in the respective portions of the subframes to reduce coding error in the sub-subframes corresponding to the pre-transient scale factors;
• and
• a multiplexer that packs and multiplexes the encoded subband signals, the transient codes and a filter code into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A multi-channel audio encoder comprising:

• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames;
• a plurality of filters that split the channels' successive data frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a plurality of predictive subband encoders each comprising a predictor and a quantizer that generate and code a difference signal for each subframe to produce encoded subband signals;
• an analyzer that splits each subframe in the audio window into a plurality of sub-subframes, creates an estimated difference signal, detects transients in the estimated difference signal in each sub-subframe, generates a transient code that indicates whether there is a transient in any sub-subframe other than the first and in which sub-subframe the transient occurs, and when a transient is detected generates a pre-transient scale factor for those sub-subframes before the transient and a post-transient scale factor for those sub-subframes including and after the transient and otherwise generates a uniform scale factor for the subframe, said analyzer further computing a transient content for the audio window based upon the transient detector in each subframe;
• a global bit manager (GBM) that uses a psychoacoustic allocation scheme to assign coding bits to each subframe in the audio window, said GBM applying a perceptual analysis window to the channels' data frames to compute a signal-to-mask ratio (SMR) for each subframe associated with the audio window when the transient content is low and allocating bits based upon the SMRs and when the transient content exceeds a transient threshold the GBM disables the psychoacoustic allocation scheme and uses a minimum mean-square-error (mmse) routine over the audio window to allocate bits to all of the subframes, said GBM assigning coding bits in said psychoacoustic allocation scheme and said mmse routine based on the estimated difference signal generated from the audio data,
• said predictive subband encoders using said pre-transient, post-transient and uniform scale factors to scale the difference signal in the respective portions of the subframes to reduce coding error in the sub-subframes corresponding to the pre-transient scale factors;
• and
• a multiplexer that packs and multiplexes the encoded subband signals and the transient codes into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A multi-channel audio encoder that encodes a multi-channel audio signal at a known bit rate, comprising:

• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames;
• a plurality of filters each comprising non-perfect and perfect reconstruction filters that split the audio frames into respective pluralities of frequency subbands over a baseband frequency range when the known bit rate is respectively below and above a threshold bit rate, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a plurality of subband encoders that code the audio data in the respective frequency bands a subframe at a time into encoded subband signals;
• and
• a multiplexer that packs and multiplexes the encoded subband signals and filter selection code into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A multi-channel audio encoder, comprising:

• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames;
• a plurality of filters that split the channels' audio frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a global bit manager (GBM) that computes a psychoacoustic signal-to-mask ratio (SMR) and an estimated prediction gain (P.sub.gain) for each subframe based upon the difference between the audio data and a predicted signal, computes mask-to-noise ratios (MNRs) by reducing the SMRs by respective fractions of their associated prediction gains, allocates bits to satisfy each MNR, computes an allocated bit rate over the subbands, and adjusts the individual allocations such that the allocated bit rate approximates a target bit rate;
• a plurality of predictive subband encoders that generate and code a difference signal in the respective frequency subbands a subframe at a time in accordance with the bit allocation to produce encoded subband signals;
• and
• a multiplexer that packs and multiplexes the encoded subband signals and bit allocation into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A multi-channel audio encoder, comprising:

• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames;
• a plurality of filters that split the channels' data frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a global bit manager (GBM) that computes a psychoacoustic signal-to-mask ratio (SMR) for each subframe based upon the difference between the audio data and a predicted signal, allocates bits to satisfy each SMR, computes an allocated bit rate over the subbands, and when the allocated bit rate is less than a target bit rate uses a minimum mean-square-error (mmse) routine to allocate the remaining bits;
• a plurality of predictive subband encoders that generate and code a difference signal in the respective frequency bands a subframe at a time in accordance with the bit allocation to produce encoded subband signals;
• and
• a multiplexer that packs and multiplexes the encoded subband signals and bit allocation into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A multi-channel fixed distortion variable rate audio encoder, comprising:

• a programmable controller for selecting one of a fixed perceptual distortion and a fixed minimum mean-square-error (mmse) distortion;
• a frame grabber that applies an audio window to each channel of a multi-channel audio signal sampled at a sampling rate to produce respective sequences of audio frames;
• a plurality of filters that split the channels' audio frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• a global bit manager (GBM) that responds to the distortion selection by selecting from an associated mmse scheme that computes a root-mean-square (RMS) value for each subframe based on the difference between the audio data and a predicted signal and allocates bits to subframes based upon the RMS values until the fixed mmse distortion is satisfied and from a psychoacoustic scheme that computes a signal-to-mask ratio (SMR) and an estimated prediction gain (P.sub.gain) for each subframe based on the difference between the audio data and a predicted signal, computes mask-to-noise ratios (MNRs) by reducing the SMRs by respective fractions of their associated prediction gains, and allocates bits to satisfy each MNR;
• a plurality of predictive subband encoders that code a difference signal derived from the audio data in the respective frequency bands a subframe at a time in accordance with the bit allocation to produce encoded subband signals;
• and
• a multiplexer that packs and multiplexes the encoded subband signals and bit allocation into an output frame for each successive data frame thereby forming a data stream at a transmission rate.

A method for encoding a multi-channel audio signal sampled at a sampling rate, comprising:

• applying an audio window to each channel of a multi-channel audio signal to produce respective sequences of audio frames;
• splitting the channels' audio frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame, each said subframe comprising at least one sub-subframe;
• encoding the audio data in the respective frequency subbands a subframe at a time into encoded subband signals;
• and
• multiplexing the encoded subband signals into an output frame for each successive audio frame to generate a data stream at a transmission rate, the size of said audio window being selected based on the ratio of the transmission rate to the sampling rate so that the size of said output frames is constrained to lie in a desired range, the size of said output frames, the number of subframes and the number of sub-subframes being multiplexed into said output frame.

A method for encoding a multi-channel audio signal sampled at a sampling rate, comprising:

• applying an audio window to each channel of a multi-channel audio signal to produce respective sequences of audio frames, said audio frames having an audio bandwidth that extends from DC to approximately half the sampling rate;
• splitting each of said audio frames into baseband frames that represent a baseband portion of the audio bandwidth and high sampling rate frames that represent the remaining portion of the audio bandwidth;
• encoding the high sampling rate frames into respective high sampling rate signals;
• splitting the channels' baseband frames into respective pluralities of frequency subbands, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame, and wherein each subframe comprises at least one sub-subframe;
• encoding the audio data in the respective frequency bands a subframe at a time into encoded subband signals;
• multiplexing the encoded subband signals and high sampling rate signals into an output frame for each successive data frame to generate a data stream at a transmission rate in which the baseband and high sampling rate portions of the multi-channel audio signal are independently decodable, the size of the audio window being set based on a ratio of the transmission rate to the sampling rate so that the size of said output frame is constrained to lie in a desired range;
• and
• multiplexing the size of said output frames, the number of subframes and the number of sub-subframes being multiplexed into said output frame.

A method for encoding a multi-channel audio signal sampled at a sampling rate and encoded at a known bit rate, comprising:

• a) applying an audio window to each channel of a multi-channel audio signal to produce respective sequences of audio frames;
• b) splitting the channels' data frames into respective pluralities of frequency subbands over a baseband frequency range by selecting a non-perfect filter bank to split the channels' audio frames when the known bit rate is below a threshold bit rate and selecting a perfect filter bank to split the channels' audio frames when the known bit rate is above the threshold bit rate, said frequency subbands each comprising a sequence of subband frames whose audio signals are subdivided into at least one subframe of audio data per subband frame;
• c) encoding the audio data in the respective frequency subbands' audio signals a subframe at a time into encoded subband signals by:
• splitting the subframe into a plurality of sub-subframes;
• detecting transients in each sub-subframe;
• generating a transient code that indicates whether there is a transient in any sub-subframe other than the first and in which sub-subframe occurs;
• generating a pre-transient scale factor for those sub-subframes before the transient and a post-transient scale factor for those sub-subframes including and after then transient when a transient is detected and otherwise generating a uniform scale factor for the subframe;
• scaling the sub-subframes in accordance with the pre-transient, post-transient and uniform scale factors;
• and
• quantizing the scaled sub-subframes at a fixed bit rate over the current subframe to generate the encoded subband signal;
• and
• d) multiplexing the encoded subband signals into an output frame for each successive data frame to generate a data stream at a transmission rate.

A method for encoding a multi-channel audio signal sampled at a sampling rate, comprising:

• applying an audio window to each channel of a multi-channel audio signal to produce respective sequences of audio frames;
• splitting the channels' data frames into respective pluralities of frequency subbands over a baseband frequency range, said frequency subbands each comprising a sequence of subband frames that have at least one subframe of audio data per subband frame;
• generating a bit allocation for the subframes in the audio window by:
• generating an estimated difference signal from said audio data for each subframe
• computing a psychoacoustic signal-to-mask ratio (SMR) for each subframe based on said estimated difference signal;
• allocating bits to satisfy each subframe's SMR;
• computing an allocated bit rate for all of the subframes;
• and
• when the allocated bit rate is less than a target bit rate, allocating the remaining bits to the subframes in accordance with a minimum mean-square-error (mmse) scheme;
• encoding a difference signal derived from the audio data in the respective frequency subbands a subframe at a time using predictive coding in accordance with the bit allocation to produce encoded subband signals;
• and
• multiplexing the encoded subband signals into an output frame for each successive data frame to generate a data stream at a transmission rate.

A method for reconstructing a multi-channel audio signal from a stream of encoded audio frames, in which each audio frame includes a sync word, a frame header, an audio header, and at least one subframe, which includes audio side information including bit allocations, a plurality of sub-subframes having baseband audio codes over a baseband frequency range, a block of high sampling rate audio codes over a high sampling rate frequency range, and an unpack sync, the method for reconstructing each audio frame comprising;

• detecting the sync word;
• unpacking the frame header to extract a frame size that indicates the number of bytes in the frame, a window size that indicates a number of audio samples in the audio frame and an encoder sampling rate;
• unpacking the audio header to extract the number of subframes and the number of audio channels;
• sequentially unpacking each subframe by:
• extracting the audio side information including the number of sub-subframes,
• demultiplexing the baseband audio codes in each sub-subframe into the multiple audio channels,
• unpacking each of the demultiplexed audio channels into a plurality of subband audio codes at respective subband frequencies,
• unpacking the high sampling rate audio codes up to a decoder sampling rate,
• skipping the remaining high sampling rate audio codes up to the encoder sampling rate, and
• detecting the unpack sync to verify the end of the subframe;
• decoding the subband audio codes in accordance with their side information to generate reconstructed subband signals a subframe at a time without reference to any other subframe;
• combining the channels' reconstructed subband signals into respective reconstructed baseband signals a subframe at a time;
• decoding the unpacked high sampling rate audio codes in accordance with their side information to generate reconstructed high sampling rate signals for each audio channel a subframe at a time;
• and
• combining the reconstructed baseband and high sampling rate signals into a reconstructed multi-channel audio signal a subframe at a time.

A method for reconstructing a multi-channel audio signal from a stream of encoded audio frames, in which each audio frame includes a sync word, a frame header. an audio header, and at least one subframe, which includes audio side information, a plurality of sub-subframes having baseband audio codes over a baseband frequency range, a block of high sampling rate audio codes over a high sampling rate frequency range, and an unpack sync, the method for reconstructing each audio frame comprising:

• detecting the sync word;
• unpacking the frame header to extract a frame size that indicates the number of bytes in the frame, a window size that indicates a number of audio samples in the audio frame, and an encoder sampling rate;
• unpacking the audio header to extract the number of subframes and the number of audio channels;
• sequentially unpacking each subframe by:
• extracting the audio side information,
• demultiplexing the baseband audio codes in each sub-subframe into the multiple audio channels,
• unpacking each of the demultiplexed audio channels into a plurality of subband audio codes at respective subband frequencies,
• unpacking the high sampling rate audio codes up to a decoder sampling rate,
• skipping the remaining high sampling rate audio codes up to the encoder sampling rate, and
• detecting the unpack sync to verify the end of the subframe;
• decoding the subband audio codes in accordance with their side information to generate reconstructed subband signals a subframe at a time without reference to any other subframe;
• combining the channels' reconstructed subband signals into respective reconstructed baseband signals a subframe at a time by unpacking the frame header to extract a reconstruction filter code, selecting a non-perfect filter bank to combine the channels' audio frames when indicated by the reconstruction filter code, and selecting a perfected filter bank to split the channels' audio frames when indicated by the reconstruction filter code;
• decoding the unpacked high sampling rate audio codes in accordance with their side information to generate reconstructed high sampling rate signals for each audio channel a subframe at a time;
• and
• combining the reconstructed baseband and high sampling rate signals into a reconstructed multi-channel audio signal a subframe at a time.