Quantization of harmonic amplitudes representing speech
AssigneeDigital Voice Systems, Inc.
InventorsJae S. Lim, John C. Hardwick
Life of this patent
Estimating the term…
Abstract
In a speech coding and decoding system, in which a timewise segment of an acoustic speech signal is represented by a frame of a data signal characterized by a fundamental frequency and spectral harmonics, a current frame is reconstructed using a set of prediction signals based on the number of spectral harmonics for the current frame and a preceding frame and reconstructed signal parameters characterizing the preceding frame. The number of spectral harmonics for the current and preceding frames are reconstructed from at least a pair of digitally encoded signals that are generated using error protection codes for all of their bits.
Claims17
A method for predicting speech signal parameters for use in encoding or decoding speech, comprising:
- representing segments of a speech signal by frames of a data signal that each comprise a fundamental frequency and multiple spectral harmonics;
- reconstructing speech signal parameters that characterize a first segment of speech from a first frame;
- reconstructing a first control parameter that characterizes the first segment using a first digitally encoded signal, wherein all digital bits of the first digitally encoded signal that represent the first control parameter are protected by an error protection code;
- reconstructing a second control parameter that characterizes a second segment of speech from a second frame using a second digitally encoded signal, wherein all digital bits of the second digitally encoded signal that represent the second control parameter are protected by an error protection code;
- and
- predicting speech signal parameters that characterize the second segment based on the reconstructed speech signal parameters and the first and second control parameters.
A method for predicting speech signal parameters for use in encoding or decoding speech, comprising:
- representing segments of a speech signal by frames of a data signal that each comprise a fundamental frequency and multiple spectral harmonics;
- reconstructing speech signal parameters that characterize a first segment of speech from a first frame;
- reconstructing a first control parameter that characterizes the first segment;
- reconstructing a second control parameter that characterizes a second segment of speech from a second frame;
- predicting speech signal parameters that characterize the second segment based on the reconstructed speech signal parameters and the first and second control parameters;
- computing a scale factor from the first and second control parameters;
- and
- applying the scale factor to the predicted speech signal parameters to produce modified predicted speech signal parameters.
A method for predicting speech signal parameters for use in encoding or decoding speech, comprising:
- representing segments of a speech signal by frames of a data signal that each comprise a fundamental frequency and multiple spectral harmonics;
- reconstructing speech signal parameters that characterize a first segment of speech from a first frame;
- reconstructing a first control parameter that characterizes the first segment;
- reconstructing a second control parameter that characterizes a second segment of speech from a second frame;
- predicting speech signal parameters that characterize the second segment based on the reconstructed speech signal parameters and the first and second control parameters;
- and
- removing the mean value of the predicted speech signal parameters from the predicted speech signal parameters to produce modified predicted speech signal parameters.
Dependencies are inferred from the claim text.