Abstract: A method of bit allocation is described. The method includes, for each among a plurality of vectors, calculating a corresponding one of a plurality of gain factors. The method also includes, for each among the plurality of vectors, calculating a corresponding bit allocation that is based on the gain factor. The method further includes, for at least one among the plurality of vectors, determining that the corresponding bit allocation is not greater than a minimum allocation value. The method additionally includes, in response to the determining, for each of the at least one vector, changing the corresponding bit allocation.

Abstract: A method of processing an audio signal is described. The method includes selecting one among a plurality of entries of a codebook based on information from the audio signal. The method also includes determining locations, in a frequency domain, of zero-valued elements of a first signal that is based on the selected codebook entry. The method further includes calculating energy of the audio signal at the determined frequency-domain locations. The method additionally includes calculating a value of a measure of a distribution of the energy of the audio signal among the determined frequency-domain locations. The method also includes calculating a noise injection gain factor based on the calculated energy and the calculated value.

Abstract: Improved audio classification is provided for encoding applications. An initial classification is performed, followed by a finer classification, to produce speech classifications and music classifications with higher accuracy and less complexity than previously available. Audio is classified as speech or music on a frame by frame basis. If the frame is classified as music by the initial classification, that frame undergoes a second, finer classification to confirm that the frame is music and not speech (e.g., speech that is tonal and/or structured that may not have been classified as speech by the initial classification). Depending on the implementation, one or more parameters may be used in the finer classification. Example parameters include voicing, modified correlation, signal activity, and long term pitch gain.

Abstract: A method of noise-robust speech classification is disclosed. Classification parameters are input to a speech classifier from external components. Internal classification parameters are generated in the speech classifier from at least one of the input parameters. A Normalized Auto-correlation Coefficient Function threshold is set. A parameter analyzer is selected according to a signal environment. A speech mode classification is determined based on a noise estimate of multiple frames of input speech.

Abstract: A scheme for coding a set of transform coefficients that represent an audio-frequency range of a signal uses a harmonic model to parameterize a relationship between the locations of regions of significant energy in the frequency domain.

Abstract: A multistage shape vector quantizer architecture uses information from a selected first-stage codebook vector to generate a rotation matrix. The rotation matrix is used to rotate the direction of the input vector to support shape quantization of the first-stage quantization error.

Abstract: A method of noise-robust speech classification is disclosed. Classification parameters are input to a speech classifier from external components. Internal classification parameters are generated in the speech classifier from at least one of the input parameters. A Normalized Auto-correlation Coefficient Function threshold is set. A parameter analyzer is selected according to a signal environment. A speech mode classification is determined based on a noise estimate of multiple frames of input speech.

Abstract: A scheme for injecting noise at uncoded elements of a spectrum is controlled according to a measure of a distribution of energy of the original spectrum among the locations of the uncoded elements.

Abstract: A dynamic bit allocation operation determines a bit allocation for each of a plurality of vectors, based on a corresponding plurality of gain factors, and compares each allocation to a threshold value that is based on a dimensionality of the vector.

Abstract: A multistage shape vector quantizer architecture uses information from a selected first-stage codebook vector to generate a rotation matrix. The rotation matrix is used to rotate the direction of the input vector to support shape quantization of the first-stage quantization error.

Abstract: A scheme for coding a set of transform coefficients that represent an audio-frequency range of a signal uses information from a reference frame that describes a previous frame of the signal to determine frequency-domain locations of regions of significant energy in a target frame of the signal.

Abstract: A scheme for coding a set of transform coefficients that represent an audio-frequency range of a signal uses a harmonic model to parameterize a relationship between the locations of regions of significant energy in the frequency domain.