Speed-variable speech signal reproduction apparatus and method

• US 5,864,792 A
• Filed: 08/12/1996
• Issued: 01/26/1999
• Est. Priority Date: 09/30/1995
• Status: Expired due to Term

• Current Assignee: Qiang Technologies LLC (Intellectual Ventures LLC)
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Kim, Chul Hong
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): RICHARDSON, SCO
• Application Number:
  US08/695,776
• Time in Patent Office: 897 Days
• Field of Search: 704/267, 704/258, 704/207, 704/201, 704/208, 704/216
• US Class Current: 704/208
• CPC Class Codes: G10L 21/04 Time compression or expansion
• CPC Class Codes: G10L 25/90 Pitch determination of speech signals

Abstract

A speed-variable speech signal reproduction apparatus and method for playing back speech signals stored in a storage medium at an adjusted speed while preventing any degradation in tone or loss of the speech signals from occurring. The method includes the steps of detecting the pitch of input digital speech signals using an average magnitude difference function, separating voice and voiceless sounds of the speech signals from each other based on the result of the detecting step, temporarily storing the separated voiceless sound, modulating the lengths of the speech signals by copying or eliminating a part of the separated voice sound, and synthesizing the modulated voice sound step with the voiceless sound temporarily stored in the storing step. The apparatus includes the a detector for detecting the pitch of input digital speech signals using an average magnitude difference function, a device for separating voice and voiceless sounds of the speech signals from each other based on the result of the detecting step, a memory for temporarily storing the separated voiceless sound, a modulator for modulating the lengths of the speech signals by copying or eliminating a part of the separated voice sound, and a synthesizer for synthesizing the modulated voice sound step with the voiceless sound temporarily stored in the storing step.

Claims

1. A speed-variable speech signal reproduction method using a signal processor adapted to receive and process digital speech signals, a memory adapted to store the digital speech signals processed by the signal processor, and a microcomputer adapted to control both the signal processor and memory, the method comprising the steps of:

(a) detecting a pitch of the digital speech signals;

(b) separating voice and voiceless sounds of the speech signals from each other based on the result of the detecting step;

(c) temporarily storing the voiceless sound separated in the separating step;

(d) modulating the lengths of the speech signals by copying or eliminating a part of the voice sound separated in the separating step; and

(e) synthesizing the voice sound modulated in the modulating step with the voiceless sound temporarily stored in the memory in the temporarily storing step;

wherein the detection of the pitch of the speech signals performed in the detecting step is achieved using the following equation;

##EQU5## where, N;

a certain segment of a window function;

m;

the sampling position;

k;

the time constant corresponding to the particular speech signal pitch to be detected.

2. A speed-variable speech signal reproduction method using a signal processor adapted to receive and process digital speech signals, a memory adapted to store the digital speech signals processed by the signal processor, and a microcomputer adapted to control both the signal processor and memory, the method comprising the steps of:

(a) detecting a pitch of the digital speech signals;

(b) separating voice and voiceless sounds of the speech signals from each other based on the result of the detecting step;

(c) temporarily storing the voiceless sound separated in the separating step;

(d) modulating the lengths of the speech signals by copying or eliminating a part of the voice sound separated in the separating step; and

(e) synthesizing the voice sound modulated in the modulating step with the voiceless sound temporarily stored in the memory in the temporarily storing step;

wherein the synthesis of the modulated voice sound with the voiceless sound carried out at the fifth step is achieved using the following equation;

##EQU6## where, α

_q ;

a variable for adjusting the amount of synthesized speech;

a modulated speech;

x(n);

a modulated speech characteristic (x(n)=x(n-δ

_q);

t_q (n);

the position of each modulated speech source; and

δ

_q ;

a variable for determining the play-back speed.

3. A speed-variable speech signal reproduction apparatus, comprising:

a detector which detects a pitch of the digital speech signals;

a separator which separates voice and voiceless sounds of the speech signals from each other based on the pitch detected by the detector;

a memory adapted to temporarily store the voiceless sound separated by the separator;

a modulator which modulates the lengths of the speech signals by copying or eliminating a part of the voice sound separated in the separating step; and

a synthesizer which synthesizes the voice sound modulated by the modulator with the voiceless sound temporarily stored in the memory;

wherein the detection of the pitch of the speech signals performed in the detector is achieved using the following equation;

##EQU7## where, N;

a certain segment of a window function;

m;

the sampling position;

k;

the time constant corresponding to the particular speech signal pitch to be detected.

4. A speed-variable speech signal reproduction apparatus, comprising:

a detector which detects a pitch of the digital speech signals;

a separator which separates voice and voiceless sounds of the speech signals from each other based on the pitch detected by the detector;

a memory adapted to temporarily store the voiceless sound separated by the separator;

a modulator which modulates the lengths of the speech signals by copying or eliminating a part of the voice sound separated in the separating step; and

a synthesizer which synthesizes the voice sound modulated by the modulator with the voiceless sound temporarily stored in the memory;

wherein the synthesis of the modulated voice sound with the voiceless sound performed by the synthesizer is achieved using the following equation;

##EQU8## where, α

_q ;

a variable for adjusting the amount of synthesized speech;

a modulated speech;

x(n);

a modulated speech characteristic (x(n)=x(n-δ

_q);

t_q (n);

the position of each modulated speech source; and

δ

_q ;

a variable for determining the play-back speed.