Speech coding method and apparatus which codes spectrum parameters and an excitation signal

• US 8,160,871 B2
• Filed: 03/31/2010
• Issued: 04/17/2012
• Est. Priority Date: 04/04/2003
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Miseki, Kimio
• Primary Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US12/751,292
• Publication Number:
  US 20100250263A1
• Time in Patent Office: 748 Days
• Field of Search: None
• US Class Current: 704/223
• CPC Class Codes: G10L 19/18 Vocoders using multiple modes

Abstract

A wideband speech coding apparatus which causes an input speech signal to be represented by spectrum parameters and an excitation signal. The apparatus includes a coding unit configured to select a plurality of pulses from given pulse position candidates, and to code the excitation signal with the selected pulses; an identification unit configured to identify whether the input speech signal is a wideband speech signal or a narrowband speech signal; and a control unit configured to control the coding unit to select a pulse position candidate having a time resolution which is set in advance in accordance with the wideband speech signal, when the identification unit identifies that the input speech signal is the wideband speech signal, and to control the coding unit to lower the time resolution of the pulse position candidate, when the identification unit identifies that the input speech signal is the narrowband speech signal.

Claims

1. A wideband speech coding apparatus which causes an input speech signal to be represented by spectrum parameters and an excitation signal, and codes the spectrum parameters and the excitation signal, the wideband speech coding apparatus comprising:

a coding unit configured to select a plurality of pulses from given pulse position candidates, and to code the excitation signal with the selected pulses;

an identification unit configured to identify whether the input speech signal is a wideband speech signal or a narrowband speech signal;

a control unit configured to control the coding unit to select a pulse position candidate having a time resolution which is set in advance in accordance with the wideband speech signal, when the identification unit identifies that the input speech signal is the wideband speech signal, and to control the coding unit to lower the time resolution of the pulse position candidate, when the identification unit identifies that the input speech signal is the narrowband speech signal;

a converter configured to convert a sampling rate of the input speech signal to a predetermined sampling rate for wideband speech coding; and

an output unit to output a result of coding by the coding unit.

2. A wideband speech coding apparatus which causes an input speech signal to be represented by spectrum parameters and an excitation signal, and codes the spectrum parameters and the excitation signal, the wideband speech coding apparatus comprising:

a coding unit configured to select a plurality of pulses from given pulse position candidates, and coding the excitation signal with the selected pulses;

an identification unit configured to identify whether a sampling rate of the input speech signal is a first sampling rate to be applied to the wideband speech signal or a second sampling rate to be applied to a narrowband speech signal;

a control unit configured to control the coding unit to use a pulse position candidate having a time resolution which is set in advance in accordance with the first sampling rate, when the identification unit identifies that the sampling rate of the input speech signal is the first sampling rate, and to control the coding unit to lower the time resolution of the pulse position candidate, when the identification unit identifies that the sampling rate of the input speech signal is the second sampling rate;

a converter configured to convert the sampling rate of the input speech signal to a predetermined sampling rate for wideband speech coding; and

an output unit to output a result of coding by the coding unit.

3. A wideband speech coding method of causing an input speech signal to be represented by spectrum parameters and an excitation signal, and then coding the spectrum parameters and the excitation signal, the wideband speech coding method comprising:

a process of identifying whether the input speech signal is a wideband speech signal or a narrowband speech signal;

a first coding process of preparing, when it is identified that the input speech signal is the wideband speech signal, pulse position candidates each having a time resolution which is set in advance in accordance with the wideband speech signal, and then coding the excitation signal with a plurality of pulses selected from the pulse position candidates;

a second coding process of lowering, when it is identified that the input speech signal is the narrowband speech signal, the time resolution of the pulse position candidate, and then coding the excitation signal with a plurality of pulses selected from the pulse position candidates;

a process of converting a sampling rate of the input speech signal to a predetermined sampling rate for wideband speech coding; and

a process of outputting a result of coding in one of the first and second coding processes which is executed.

4. A wideband speech coding method of causing an input speech signal to be represented by spectrum parameters and an excitation signal, and then coding the spectrum parameters and the excitation signal, the wideband speech coding method comprising:

a process of identifying whether a sampling rate of the input speech signal is a first sampling rate to be applied to the wideband speech signal or a second sampling rate to be applied to a narrowband speech signal;

a process of preparing, when it is identified that the sampling rate of the input speech signal is the first sampling rate, pulse position candidates each having a time resolution which is set in advance in accordance with the first sampling rate, and then coding the excitation signal with a plurality of pulses selected from the first pulse position candidates;

a process of lowering, when it is identified that the sampling rate of the input speech signal is the second sampling rate, the time resolution of the pulse position candidate, and coding the excitation signal with a plurality of pulses selected from the second pulse position candidates;

a process of converting the sampling rate of the input speech signal to a predetermined sampling rate for wideband speech coding; and

a process of outputting a result of coding in one of the preparing and coding process and the lowering and coding process which is executed.

5. A wideband speech coding apparatus which causes an input speech signal to be represented by spectrum parameters and an excitation signal, and codes the spectrum parameters and the excitation signal, the wideband speech coding apparatus comprising:

an adaptive codebook searcher configured to produce a first codevector corresponding to a pitch period of the input speech signal by using an adaptive codebook;

a noise codebook searcher configured to produce a second codevector by using a noise codebook comprising arrangement information regarding pulse position candidates, pulse polarity and the number of pulses;

an excitation signal producer configured to produce the excitation signal by using the first codevector and the second codevector, and store the excitation signal in the adaptive codebook;

an output configured to output a code of the spectrum parameters coded by a spectrum parameter coder, a code corresponding to the first codevector produced by the adaptive codebook searcher, and a code corresponding to the second codevector produced by the noise codebook searcher; and

an identifier configured to identify whether the input speech signal is a wideband speech signal or a narrowband speech signal,

wherein the noise codebook searcher modifies arrangement information of the noise codebook, such that when the identifier identifies that the input speech signal is the wideband speech signal, the noise codebook searcher produces the second codevector having a first number of pulses, and when the identifier identifies that the input speech signal is the narrowband speech signal, the noise codebook searcher produces the second codevector having a second number of pulses which is larger than the first number of pulses.

6. A wideband speech coding method which causes an input speech signal to be represented by spectrum parameters and an excitation signal, and codes the spectrum parameters and the excitation signal, the wideband speech coding method comprising:

a spectrum parameters coding process of extracting spectrum parameters from the input speech signal, and coding the extracted spectrum parameters;

an adaptive codebook searching process of producing a first codevector corresponding to a pitch period of the input speech signal by using an adaptive codebook;

a noise codebook searching process of producing a second codevector by using a noise codebook comprising arrangement information regarding pulse position candidates, pulse polarity and the number of pulses;

a producing process of producing the excitation signal by using the first codevector and the second codevector;

a storing process of storing the excitation signal in the adaptive codebook;

an outputting process of outputting a code of the spectrum parameters coded in the spectrum parameters coding process, a code corresponding to the first codevector produced in the adaptive code searching process, and a code corresponding to the second codevector produced in the noise code book searching process; and

an identification process of identifying whether the input speech signal is a wideband speech signal or a narrowband speech signal,

wherein in the noise codebook searching process, arrangement information of the noise codebook is modified, such that when it is identified in the identification process that the input speech signal is the wideband speech signal, the second codevector is produced in such a manner as to have a first number of pulses, and when it is identified in the identification process that the input speech signal is the narrowband speech signal, the second codevector having a second number of pulses is produced, the second number of pulses being larger than the first number of pulses.