Speech gain quantization strategy

• US 10,181,327 B2
• Filed: 03/06/2009
• Issued: 01/15/2019
• Est. Priority Date: 05/19/2000
• Status: Expired due to Fees

• Current Assignee: DigiMedia Tech, LLC (IP Investments Group LLC)
• Original Assignee: Nytell Software LLC (Intellectual Ventures LLC)
• Inventors: Gao, Yang, Benyassine, Adil
• Primary Examiner(s): Opsasnick, Michael N
• Application Number:
  US12/381,036
• Publication Number:
  US 20090177464A1
• Time in Patent Office: 3,602 Days
• Field of Search: 704219-223
• CPC Class Codes: G10L 19/00 Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis in musical instruments G10H
• CPC Class Codes: G10L 19/083 the excitation function being an excitation gain G10L25/90 takes precedence
• CPC Class Codes: G10L 19/167 Audio streaming, i.e. formatting and decoding of an encoded audio signal representation into a data stream for transmission or storage purposes
• CPC Class Codes: G10L 19/18 Vocoders using multiple modes
• CPC Class Codes: G10L 19/24 Variable rate codecs, e.g. for generating different qualities using a scalable representation such as hierarchical encoding or layered encoding
• CPC Class Codes: H03G 3/00 Gain control in amplifiers or frequency changers

Abstract

A speech encoder that analyzes and classifies each frame of speech as being periodic-like speech or non-periodic like speech where the speech encoder performs a different gain quantization process depending if the speech is periodic or not. If the speech is periodic, the improved speech encoder obtains the pitch gains from the unquantized weighted speech signal and performs a pre-vector quantization of the adaptive codebook gain G_P for each subframe of the frame before subframe processing begins and a closed-loop delayed decision vector quantization of the fixed codebook gain G_C. If the frame of speech is non-periodic, the speech encoder may use any known method of gain quantization. The result of quantizing gains of periodic speech in this manner results in a reduction of the number of bits required to represent the quantized gain information and for periodic speech, the ability to use the quantized pitch gain for the current subframe to search the fixed codebook for the fixed codebook excitation vector for the current subframe. Alternatively, the new gain quantization process which was used only for periodic signals may be extended to non-periodic signals as well. This second strategy results in a slightly higher bit rate than that for periodic signals that use the new gain quantization strategy, but is still lower than the prior art'"'"'s bit rate. Yet another alternative is to use the new gain quantization process for all speech signals without distinguishing between periodic and non-periodic signals.

Claims

1. A speech encoding system comprising:

at least one processing circuit; and

a memory storing a first vector quantization codebook and a second vector quantization codebook, and instructions executable by the at least one processing circuit to cause the speech encoding system to transform an input speech signal into an output bitstream by;

generating a plurality of subframes of a frame of the input speech signal;

generating unquantized pitch gains for the respective subframes;

selecting a mode from a plurality of modes based on the frame of the input speech signal;

in response to the mode being a first of the modes;

selecting a quantized pitch gain vector from the first vector quantization codebook based on the unquantized pitch gains, wherein the quantized pitch gain vector is selected in an open loop fashion, wherein the quantized pitch gain vector includes adaptive-codebook gain components corresponding respectively to the subframes of the frame;

performing subframe processing based on data including the quantized pitch gain vector to select a first excitation from a fixed excitation codebook; and

selecting a second gain vector from the second vector quantization codebook using the first excitation and a closed loop minimization, wherein the second gain vector includes fixed-codebook gain components corresponding respectively to the subframes of the frame;

generating the output bitstream that includes first information indicating the quantized pitch gain vector and second information indicating the second gain vector in response to the mode being the first mode; and

transmission circuitry operable to transmit the output bitstream via a radiofrequency link, wherein the output bitstream includes information that is decodable by a remote handheld device to play a reproduced version of the input speech signal.

2. The system of claim 1, wherein the first mode is selected based on a periodicity of the input speech signal being greater than a selected value.

3. The system of claim 1, wherein the first vector quantization codebook is a three-dimensional gain codebook.

4. The system of claim 1, wherein the second vector quantization codebook is an n-dimensional gain codebook, with n greater than or equal to two.

5. The system of claim 4, wherein the n-dimensional gain codebook is a three-dimensional gain codebook.

15. The system of claim 1, wherein the selecting the quantized pitch gain vector from the first vector quantization codebook is based only on the unquantized pitch gains.

6. A method comprising:

a speech encoding system generating a plurality of subframes of a frame of an input speech signal;

the speech encoding system generating unquantized pitch gains for the respective subframes;

the speech encoding system selecting a mode from a plurality of modes based on the frame of the input speech signal;

in response to the mode being a first of the modes, the speech encoding system;

selecting a quantized pitch gain vector from a first vector quantization codebook based on the unquantized pitch gains, wherein the quantized pitch gain vector is selected in an open loop fashion, wherein the quantized pitch gain vector includes adaptive-codebook gain components corresponding respectively to the subframes of the frame;

performing subframe processing based on data including the quantized pitch gain vector to select a first excitation from a fixed excitation codebook; and

selecting a second gain vector from a second vector quantization codebook using the first excitation and a closed loop minimization, wherein the second gain vector includes fixed-codebook gain components correspond respectively to the subframes of the frame;

the speech encoding system generating an output bitstream including first information indicating the quantized pitch gain vector and second information indicating the second gain vector in response to the mode being the first mode, wherein the output bitstream is a transformed version of the input speech signal; and

transmission circuity transmitting the output bitstream via a radiofrequency link, wherein the output bitstream includes information that is decodable by a remote handheld device to play a reproduced version of the input speech signal.

7. The method of claim 6, wherein the first mode is selected based on a periodicity of the input speech signal being greater than a selected value.

8. The method of claim 6, wherein the first vector quantization codebook is a three-dimensional gain codebook.

9. The method of claim 6, wherein the second vector quantization codebook is an n-dimensional gain codebook, with n greater than or equal to two.

10. The method of claim 9, wherein the n-dimensional gain codebook is a three-dimensional gain codebook.

11. A cell phone comprising:

a receiver configured to receive a frame of encoded speech from a remote transmitter via a radiofrequency link, wherein the frame represents a portion of a speech signal;

a mode extractor configured to extract a mode from the frame, wherein the mode is one of a plurality of modes;

a gain extractor configured to determine a quantized pitch gain vector and a second gain vector based on information contained in the frame in response to the mode being a first of the modes, wherein the quantized pitch gain vector includes adaptive-codebook gain components corresponding respectively to subframes of the frame, wherein the quantized pitch gain vector has been selected prior to subframe processing from a first vector quantization codebook based on unquantized pitch gains for the respective subframes, wherein the second gain vector includes fixed-codebook gain components corresponding respectively to the subframes of the frame, wherein the second gain vector has been selected after the subframe processing from a second vector quantization codebook based on a fixed codebook excitation and a closed loop minimization;

a reconstructor configured to reconstruct the portion of the speech signal based on data including the quantized pitch gain vector and the second gain vector in response to the mode being the first mode; and

a speaker configured to play the portion of the speech signal.

12. The cell phone of claim 11, wherein the first mode corresponds to a periodicity of the speech signal being greater than a selected value.

13. A method comprising:

a receiver receiving a frame of encoded speech from a remote transmitter via a radiofrequency link, wherein the frame represents a portion of a speech signal;

a speech decoding system extracting a mode associated with the frame, wherein the mode is one of a plurality of modes;

the speech decoding system determining a quantized pitch gain vector and a second gain vector based on information contained in the frame in response to the mode being a first of the modes, wherein the quantized pitch gain vector includes adaptive-codebook gain components corresponding respectively to subframes of the frame, wherein the quantized pitch gain vector has been selected by an encoder prior to subframe processing at the encoder, wherein the quantized pitch gain vector has been selected from a first vector quantization codebook based on unquantized pitch gains for the respective subframes, wherein the second gain vector includes fixed-codebook gain components corresponding respectively to the subframes of the frame, wherein the second gain vector has been selected by the encoder after the subframe processing, wherein the second gain vector has been selected from a second vector quantization codebook based on a fixed codebook excitation and a closed loop minimization;

the speech decoding system reconstructing the portion of the speech signal based on data including the quantized pitch gain vector and the second gain vector in response to the mode being the first mode; and

a speaker playing the portion of the speech signal.

14. The method of claim 13, wherein the first mode corresponds to a periodicity of the speech signal being greater than a selected value.