A method for partitioning of input vectors for coding is presented. The method comprises obtaining of an input vector. The input vector is segmented, in a non-recursive manner, into an integer number, N.sup.SEG, of input vector segments. A representation of a respective relative energy difference between parts of the input vector on each side of each boundary between the input vector segments is determined, in a recursive manner. The input vector segments and the representations of the relative energy differences are provided for individual coding. Partitioning units and computer programs for partitioning of input vectors for coding, as well as positional encoders, are presented.

Methods, an encoder and a decoder are configured for transition between frames with different internal sampling rates. Linear predictive (LP) filter parameters are converted from a sampling rate S1 to a sampling rate S2. A power spectrum of a LP synthesis filter is computed, at the sampling rate S1, using the LP filter parameters. The power spectrum of the LP synthesis filter is modified to convert it from the sampling rate S1 to the sampling rate S2. The modified power spectrum of the LP synthesis filter is inverse transformed to determine autocorrelations of the LP synthesis filter at the sampling rate S2. The autocorrelations are used to compute the LP filter parameters at the sampling rate S2.

Disclosed are some examples of systems, apparatus, methods and computer program products implementing techniques for extending the range of a set of decoded parameter values for a sequence of frequency bands in an identifiable time frame of an audio signal. In some implementations, the parameter values vary in relation to a sequence of time frames of the audio signal and in relation to a sequence of frequency bands in each time frame. In some implementations, it is determined that a decoded value corresponds to a minimum of a first range of values of a first coding protocol of a set of coding protocols. The determined value is modified to be below the minimum of the first range of values to produce an extended value. A modified set of decoded values including one or more extended values can thus be provided.

Methods, an encoder and a decoder are configured for transition between frames with different internal sampling rates. Linear predictive (LP) filter parameters are converted from a sampling rate S1 to a sampling rate S2. A power spectrum of a LP synthesis filter is computed, at the sampling rate S1, using the LP filter parameters. The power spectrum of the LP synthesis filter is modified to convert it from the sampling rate S1 to the sampling rate S2. The modified power spectrum of the LP synthesis filter is inverse transformed to determine autocorrelations of the LP synthesis filter at the sampling rate S2. The autocorrelations are used to compute the LP filter parameters at the sampling rate S2.

Aspects of the present disclosure provide improved techniques for coding audio signal with a transient audio sound. Improved techniques include parsing a frame of predetermined length of audio samples into a series of windows of a smaller size, and transforming the windows of time-domain samples into a series of windows of frequency-domain samples. In an aspect coding of the frequency-domain samples may include vector quantization of vectors formed of frequency-domain samples selected from across the frame.

A quantization device includes: a trellis-structured vector quantizer which quantizes a first error vector between an N-dimensional (here, N is two or more) subvector and a first predictive vector; and an inter-frame predictor which generates a first predictive vector from the quantized N-dimensional subvector, wherein the inter-frame predictor uses a predictive coefficient comprising an NN matrix and performs an inter-frame prediction using the quantized N-dimensional subvector of a previous stage.

A method for partitioning of input vectors for coding is presented. The method comprises obtaining of an input vector. The input vector is segmented, in a non-recursive manner, into an integer number, N.sup.SEG, of input vector segments. A representation of a respective relative energy difference between parts of the input vector on each side of each boundary between the input vector segments is determined, in a recursive manner. The input vector segments and the representations of the relative energy differences are provided for individual coding. Partitioning units and computer programs for partitioning of input vectors for coding, as well as positional encoders, are presented.

This disclosure discloses an audio encoding/decoding method and apparatus, a storage medium, and a computer program product, and belongs to the audio encoding/decoding field. In this solution, when an audio signal is a dual-channel signal, even if a bitstream includes a left-channel bitstream and a right-channel bitstream, based on a channel decoding mode, the left-channel bitstream is decoded but the right-channel bitstream is not decoded, or the right-channel bitstream is decoded but the left-channel bitstream is not decoded in a decoding process. Therefore, power consumption at a decoder side is reduced when resources at the decoder side are limited. Correspondingly, an encoder side can also sequentially encode left-channel data and right-channel data based on a condition met by the audio signal, rather than necessarily performing encoding in a dual-channel interleaving encoding scheme or a dual-channel deinterleaving encoding scheme.

A method for partitioning of input vectors for coding is presented. The method comprises obtaining of an input vector. The input vector is segmented, in a non-recursive manner, into an integer number, N.sup.SEG, of input vector segments. A representation of a respective relative energy difference between parts of the input vector on each side of each boundary between the input vector segments is determined, in a recursive manner. The input vector segments and the representations of the relative energy differences are provided for individual coding. Partitioning units and computer programs for partitioning of input vectors for coding, as well as positional encoders, are presented.

Self-supervised mechanisms to evaluate speech quality, and self-supervised speech enhancement, based on the quantization error of a vector-quantized variational autoencoder that utilize clean speech with domain knowledge of speech processing incorporated into the model design to improve correlation with real quality scores; and a self-distillation mechanism combined with adversarial training.