Abstract: The present disclosure provides embodiments of method, system and devices to convert whispered speech to normal speech intelligently through deep learning, therefore whispered speech can be more robust to interference and more intelligible to listener. A first audio signal is received including first whispered speech; a first plurality of computations are performed on the first audio signal to extract a first features; the first features are provided as input to a trained deep neural network (DNN) model to obtain output of the DNN model comprising a second features; and an inverse of the first plurality of computations are performed on the second features to produce a second audio signal corresponding to a non-whispered version of the first whispered speech.

Abstract: The present disclosure provides embodiments of method, system and devices to convert whispered speech to normal speech intelligently through deep learning, therefore whispered speech can be more robust to interference and more intelligible to listener. A first audio signal is received including first whispered speech; a first plurality of computations are performed on the first audio signal to extract a first features; the first features are provided as input to a trained deep neural network (DNN) model to obtain output of the DNN model comprising a second features; and an inverse of the first plurality of computations are performed on the second features to produce a second audio signal corresponding to a non-whispered version of the first whispered speech.

Abstract: A method of updating the state of a decoder that decodes a wide-band signal including a plurality of sub-band signals, including: receiving the plurality of sub-band signals; for each sub-band signal, storing portions of that sub-band signal in a respective buffer; responsive to determining that a portion of the wide-band signal is degraded, performing a packet loss concealment algorithm to determine wide-band replacement data for the degraded portion; selecting a portion of the sub-band signal stored in each buffer in dependence on the determined wide-band replacement data; and updating the state of the decoder using the selected portions.

Abstract: Methods and apparatus for audio and speech processing including generating a plurality of frames, each of the frames comprising a plurality of transform coefficients, and allocating bits to the transform coefficients in each of the frames such that at least two of the transform coefficients in the same frame have different bit allocations and the total number of the bits allocated to the transform coefficients in at least two of the frames is equal.

Abstract: A method of updating the state of a decoder that decodes a wide-band signal including a plurality of sub-band signals, including: receiving the plurality of sub-band signals; for each sub-band signal, storing portions of that sub-band signal in a respective buffer; responsive to determining that a portion of the wide-band signal is degraded, performing a packet loss concealment algorithm to determine wide-band replacement data for the degraded portion; selecting a portion of the sub-band signal stored in each buffer in dependence on the determined wide-band replacement data; and updating the state of the decoder using the selected portions.

Abstract: The speech feature extraction algorithm is based on a hierarchical combination of auditory similarity and pooling functions. Computationally efficient features referred to as “Sparse Auditory Reproducing Kernel” (SPARK) coefficients are extracted under the hypothesis that the noise-robust information in speech signal is embedded in a reproducing kernel Hilbert space (RKHS) spanned by overcomplete, nonlinear, and time-shifted gammatone basis functions. The feature extraction algorithm first involves computing kernel based similarity between the speech signal and the time-shifted gammatone functions, followed by feature pruning using a simple pooling technique (“MAX” operation). Different hyper-parameters and kernel functions may be used to enhance the performance of a SPARK based speech recognizer.

Abstract: Methods and apparatus for audio and speech processing including generating a plurality of frames, each of the frames comprising a plurality of transform coefficients, and allocating bits to the transform coefficients in each of the frames such that at least two of the transform coefficients in the same frame have different bit allocations and the total number of the bits allocated to the transform coefficients in at least two of the frames is equal.