Voice signal coding apparatus

• US 6,272,459 B1
• Filed: 04/11/1997
• Issued: 08/07/2001
• Est. Priority Date: 04/12/1996
• Status: Expired due to Term

• Current Assignee: Olympus Optical Company Ltd. (Olympus Corporation)
• Original Assignee: Olympus Optical Company Ltd. (Olympus Corporation)
• Inventors: Takahashi, Hidetaka
• Primary Examiner(s): Zele, Krista
• Assistant Examiner(s): Opsasnick, Michael N.
• Application Number:
  US08/845,543
• Time in Patent Office: 1,579 Days
• Field of Search: 704/213, 704/214, 704/225, 704/226, 704/228
• US Class Current: 704/221
• CPC Class Codes: G06T 9/004 Predictors, e.g. intraframe, interframe coding
• CPC Class Codes: G10L 19/012 Comfort noise or silence coding
• CPC Class Codes: G10L 19/16 Vocoder architecture
• CPC Class Codes: G10L 25/78 Detection of presence or absence of voice signals switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10
• CPC Class Codes: G10L 25/93 Discriminating between voiced and unvoiced parts of speech signals G10L25/90 takes precedence

Abstract

A voice signal coding apparatus includes: a voice status detector detecting whether an input signal divided at predetermined frame intervals is a voice or a non-voice signal; a linear predictive analyzer outputting a linear predictive parameter associated with the input signal; a voice sound source predicting circuit; a non-voice sound source predicting circuit including a random signal generator; and a switch controller selecting either the voice or non-voice sound source predicting circuit from the detection result of the voice status detector, wherein the random signal gain is set in accordance with a value obtained by suppressing by a predetermined factor the gain obtained when a non-voice input signal is coded by the voice sound predicting circuit. When the input signal is a non-voice signal over a number of successive frames, a switch controller continuously outputs a linear predictive parameter for a previous frame; when a predetermined number of successive non-voice signal frames occurs, the parameter smoothing circuit calculates the weighted sum of the linear predictive parameter of the current frame and a linear predictive parameter used in non-voice frames and outputs a smoothed parameter; and a parameter memory stores the smoothed parameter. The voice status detector includes a threshold value adapting circuit increasing threshold value when the input signal level is greater than the current threshold value and decreasing it when the input signal level is smaller than the current threshold value, wherein the threshold value increasing factor is set to a value greater than a factor by which the threshold value is increased in a normal state.

Claims

1. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input voice signal divided into predetermined frame intervals, is a voice signal or a non-voice signal during each frame interval, a detection that each frame interval is voiced or non-voiced being based on a frame energy of each frame interval;

a linear predictive analyzer for outputting a linear predictive parameter associated with said input signal;

a linear predictive coding circuit for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coding circuit including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal; and

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coding circuit performs a coding operation, in accordance with a result given by said voice status detector.

2. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input voice signal divided into predetermined frame intervals, is a voice signal or a non-voice signal during each frame interval, a detection that each frame interval is voiced or non-voiced being based on a frame energy of each frame interval;

a linear predictive analyzer for outputting a linear predictive parameter associated with said input signal;

a linear predictive coding circuit for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coding circuit including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal; and

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coding circuit performs a coding operation, in accordance with a result given by said voice status detector;

said non-voice sound source predicting part including a random signal generator for generating a random signal representing a sound source signal;

data coded by said non-voice sound source predicting part including said linear predictive parameter, said random signal, and a gain associated with said random signal; and

said gain associated with said random signal being set in accordance with a gain obtained when said voice sound source predicting part codes a non-voice signal given as the input signal.

3. A voice signal coding apparatus according to claim 2, wherein said gain associated with the random signal is set in accordance with a value obtained by suppressing, by a predetermined factor, a gain obtained when said voice sound source predicting part codes said non-voice signal given as the input signal.

4. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input voice signal divided into predetermined frame intervals, is a voice signal or a non-voice signal during each frame interval, a detection that each frame interval is voiced or non-voiced being based on a frame energy of each frame interval;

a linear predictive analyzer for outputting a linear predictive parameter associated with said input signal;

a linear predictive coding circuit for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coding circuit including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal;

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coding circuit performs a coding operation, in accordance with a result given by said voice status detector;

a controller for controlling the linear predictive parameter such that when the detection result by said voice status detector indicates that the input signal is a non-voice signal over a predetermined number of successive frames, the linear predictive parameter employed for a previous frame is continuously employed as the linear predictive parameter for the input signal;

a smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames in such a manner that when a number of successive frames regarded as including a non-voice signal exceeds said predetermined number, smoothing is performed between a previously-prepared linear predictive parameter for use in non-voice frames and a linear predictive parameter associated with the current frame, and the resultant smoothed linear predictive parameter is output.

5. A voice signal coding apparatus according to claim 4, wherein a value of the linear predictive parameter predetermined for particular background noise is employed as an initial value of said linear predictive parameter for use in non-voice frames.

6. A voice signal coding apparatus according to claim 4, wherein

said smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames performs smoothing with a greater weighting factor for said linear predictive parameters for use in non-voice frames than that for the linear predictive parameter for the current frame.

7. A voice signal coding apparatus according to claim 5, wherein

said smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames performs smoothing with a greater weighting factor for said linear predictive parameter for use in non-voice frames than that for the linear predictive parameter for the current frame.

8. A voice signal coding apparatus according to claim 4, further comprising:

a parameter storing device for storing a linear predictive parameter output by said smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames so that said linear predictive parameter stored may be used as a linear predictive parameter for use in non-voice frames to smooth a next frame.

9. A voice signal coding apparatus according to claim 5, further comprising:

a parameter storing device for storing a linear predictive parameter output by said smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames so that said linear predictive parameter stored may be used as a linear predictive parameter for use in non-voice frames to smooth a next frame.

10. A voice signal coding apparatus according to claim 6, further comprising:

a parameter storing circuit for storing a linear predictive parameter output by said smoothing circuit for smoothing the linear predictive parameter for use in non-voice frames so that said linear predictive parameter stored may be used as a linear predictive parameter for use in non-voice frames to smooth a next frame.

11. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input voice signal divided into predetermined frame intervals, is a voice signal or a non-voice signal during each frame interval, a detection that each frame interval is voiced or non-voiced being based on a frame energy of each frame interval;

a linear predictive analyzer for outputting a linear predictive parameter associated with said input signal;

a linear predictive coding circuit for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coding circuit including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal;

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coding circuit performs a coding operation, in accordance with a result given by said voice status detector;

said voice signal status detector comprising;

a level measuring circuit for measuring a level of an input signal divided at predetermined frame intervals;

a comparator for comparing the level of said input signal measured by said level measuring circuit with a threshold value provided as a criterion for determining whether said input signal is a voice signal or a non-voice signal and then outputting the comparison result; and

a threshold value adapting circuit by which said threshold value is increased by a predetermined factor when said comparator has determined that the level of said input signal is greater than a current threshold value, while said threshold value is decreased when said comparator has determined that the level of said input signal is smaller than the current threshold value, wherein said predetermined factor by which said threshold value is increased is set in accordance with a difference between the current threshold value and a minimum level of the input signal during a predetermined period of past time.

12. A voice signal coding apparatus according to claim 11, wherein when the level of said input signal is regarded by said comparator as smaller than the current threshold value, and said threshold value adapting circuit employs said level of said input signal as a new threshold value.

13. A voice signal coding apparatus according to claim 11, wherein

when a current threshold value is compared with a minimum level of the input signal during a predetermined period of time from a start of a voice status detecting process, if said current threshold value is smaller than said minimum level by an amount equal to or greater than a predetermined value, said threshold value adapting circuit sets said factor, by which the threshold value is increased, to a predetermined value which is greater than a factor by which the threshold value is increased in a normal state, and resets said factor, by which the threshold value is increased, to said normal value when said input signal is regarded as a non-voice signal.

14. A voice signal coding apparatus according to claim 12, wherein

when a current threshold value is compared with a minimum level of the input signal during a predetermined period of time from a start of a voice status detecting process, if said current threshold value is smaller than said minimum level by an amount equal to or greater than a predetermined value, said threshold value adapting circuit sets said factor, by which the threshold value is increased, to a predetermined value which is greater than a factor by which the threshold value is increased in a normal state, and resets said factor, by which the threshold value is increased, to said normal value when said input signal is regarded as a non-voice signal.

15. A method for coding voice signals comprising the steps of:

receiving an input voice signal comprised of voice and non-voice signal components;

dividing the receiving input signal into predetermined frames of a given interval;

examining each frame to determine if it is a voice or non-voice signal; and

respectively coding one of a voice signal and a non-voice signal for each frame responsive to identification of said frame as having a voice or non-voice signal;

said non-voice signal including a random signal which is employed as a sound source signal; and

selecting a predicted sound source signal from probability code vectors stored in a probability code book, determining an energy in the predicted sound source signal selected and modifying the random signal by adjusting a gain of the random signal until an energy of the generated random signal is substantially equal to the energy of the predicted sound source signal.

16. A method for coding voice signals comprising the steps of:

receiving an input voice signal comprised of voice and non-voice signal components;

dividing the receiving input signal into predetermined frames of a given interval;

examining each frame to determine if it is a voice or non-voice signal;

respectively coding one ofa voice signal and a non-voice signal for each frame responsive to identification of said frame as having a voice or non-voice signal;

said non-voice signal including a random signal which is employed as a sound source signal; and

selecting a predicted sound source signal from probability code vectors stored in a probability code book, determining an energy in the predicted sound source signal selected and modifying the random signal by adjusting a gain of the random signal until an energy of the generated random signal is substantially equal to the energy ofthe predicted sound source signal; and

generating a linear predictive parameter representing a spectrum characteristic of the input signal;

modifying the coded non-voice signal by said parameter subtracting the modified coded non-voice signal from a sub-frame of the input signal to generate an error vector;

weighting the error vector according to audibility characteristics; and

calculating a mean square ofthe error vector which is employed to select an optimum code stored in the probability code book.

17. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input signal comprised of voice and non-voice components and divided at predetermined frame intervals, is a voice signal or a non-voice signal;

a linear predictive analyzer outputting a linear predictive parameter associated with said input signal;

a linear predictive coder for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coder including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal;

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coder performs a coding operation, in accordance with a result given by said voice status detector;

said non-voice sound source predicting part including a random signal generator for generating a random signal representing a sound source signal;

a circuit for adjusting a gain of the random signal employing an output from the voice sound source predicting part;

a device employing the linear predictive parameter to modify the input signal;

a subtractor subtracting the modified input signal from the original input signal; and

a circuit for evaluating the result of the subtracting circuit to modify a coded signal from one of said voice and non-voice sound signal predicting parts.

18. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input voice signal divided into predetermined frame intervals, is a voice signal or a non-voice signal during each frame interval, a detection that each frame interval is voiced or non-voiced being based on a frame energy of each frame interval;

a linear predictive analyzer for outputting a linear predictive parameter associated with said input signal;

a linear predictive coding circuit for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coding circuit including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal;

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coding circuit performs a coding operation, in accordance with a result given by said voice status detector.

said voice signal status detector comprising;

a first frame analyzer for determining an energy of an initial frame;

a threshold decision circuit for determining a threshold value based on an output of the first frame analyzer;

a second frame analyzer for detecting energy of at least one frame subsequent to the initial frame;

a third circuit for subtracting the initial frame energy from the output of said second frame analyzer; and

a detector for comparing a difference value outputted by the the third circuit with the threshhold value generated by said threshold decision circuit to determine whether an examined frame is voice or non-voice, said determination being available for a frame immediately following the initial frame.

19. A method for coding voice signals comprising the steps of:

(a) receiving an input voice signal comprised of voice and non-voice signal components;

(b) dividing the receiving input signal into predetermined frames of a given interval;

(c) examining each frame to determine if it is a voice or non-voice signal;

(d) respectively coding one of a voice signal and a non-voice signal for each frame responsive to identification of said frame as having a voice or non-voice signal;

said non-voice signal including a random signal which is employed as a sound source signal; and

(e) selecting a predicted sound source signal from probability code vectors stored in a probability code book, determining an energy in the predicted sound source signal selected and modifying the random signal by adjusting a gain of the random signal until an energy of the generated random signal is substantially equal to the energy ofthe predicted sound source signal;

(f) step (c) further comprising measuring a level of an input signal of each frame;

(g) comparing the level of said input signal at step (f) with a threshold value provided as a criterion for determining whether said input signal is a voice signal or a non-voice signal and then outputting the comparison result; and

(h) increasing said threshold value by a predetermined factor when the comparing step (g) determines that the level of said input signal is greater than a current threshold value, and decreasing said threshold value when said comparing step (g) determines that the level of said input signal is smaller than the current threshold value, wherein said predetermined factor by which said threshold value is changed is set in accordance with a difference between the current threshold value and a minimum level of the input signal during a predetermined period of past time.

20. A voice signal coding apparatus comprising:

a voice status detector for detecting whether an input signal divided at predetermined frame intervals is a voice signal or a non-voice signal;

a linear predictive analyzer outputting a linear predictive parameter associated with said input signal;

a linear predictive coder for generating a driving sound source signal corresponding to a residual linear predictive signal, said linear predictive coder including a voice sound source predicting part for coding a voice signal and a non-voice sound source predicting part for coding a non-voice signal;

a coding mode selector for selecting either a voice signal coding mode or a non-voice signal coding mode in which said linear predictive coder performs a coding operation, in accordance with a result given by said voice status detector;

said non-voice sound source predicting part including a random signal generator for generating a random signal representing a sound source signal;

a circuit for adjusting a gain of the random signal employing an output from the voice sound source predicting part;

a device employing the linear predictive parameter to modify the input signal;

a subtractor subtracting the modified input signal from the original input signal; and

a circuit for evaluating the result of the subtracting circuit to modify a coded signal from one of said voice and non-voice sound signal predicting parts; and

a voice signal status detector comprising;

a level measuring circuit for measuring a level of an input signal divided at predetermined frame intervals;

a comparator for comparing the level of said input signal measured by said level measuring circuit with a threshold value provided as a criterion for determining whether said input signal is a voice signal or a non-voice signal and then outputting the comparison result; and

a threshold value adapting circuit by which said threshold value is increased by a predetermined factor when said comparator has determined that the level of said input signal is greater than a current threshold value, while said threshold value is decreased when said comparator has determined that the level of said input signal is smaller than the current threshold value, wherein said predetermined factor by which said threshold value is increased is set in accordance with a difference between the current threshold value and a minimum level of the input signal during a predetermined period of past time.