« PreviousContinue »
United States Patent  [ii] Patent Number: 5,668,927
Chan et al.  Date of Patent: Sep. 16, 1997
 METHOD FOR REDUCING NOISE IN SPEECH SIGNALS BY ADAPTTVELY CONTROLLING A MAXIMUM LIKELIHOOD FILTER FOR CALCULATING SPEECH COMPONENTS
 Inventors: Joseph Chan. Tokyo; Masayuki
NLshiguchi. Kanagawa, both of Japan
 Assignee: Sony Corporation, Tokyo, Japan
 Appl. No.: 431,746
 Filed: May 1,1995
 Foreign Application Priority Data
May 13, 1994 [JP] Japan 6-099869
 IntCl.6 G10L5/06
 U.S. CI 704/240; 704/214
 Field of Search 395/2.09, 2.1,
395/2.14, 2.17, 2.19, 2.2. 2.23. 2.24, 2.25, 2.35, 2.36, 2.37, 2.4, 2.42. 2.45, 2.46, 2.48,
 References Cited
U.S. PATENT DOCUMENTS
5,036540 7/1991 Eastmond 395/2.35
T. Parsons. "Voice and Speech Processing," 1987, pp.
170-175, 219-222, 345-353, 362.
J.R. Deller et al., "Discrete-Time Processing of Speech
Signals," 1987, pp. 506-516.
S. Furui, "Digital Speech Processing, Synthesis, and Recognition," 1989, pp. 91-98.
L.R. Rabiner, "Digital Processing of Speech Signals," 1978, pp. 158-161.
Boll, "Suppression of Acoustic Noise in Speech Using Spectral Subtraction," IEEE Trans, on Acoustics, Speech, and Signal Processing, 27(2): 113-120. Apr. 1979.
G. Whipple, "Low Residual Noise Speech Enhancement Utilizing Time-Frequency Filtering." ICASSP-94, Apr. 19-22, 1994, pp. 5-8.
M. Nishiguchi, "Vector Quantized MBE with Simplified V/UV Division at 3.0 kbps," ICASSP-93, Apr. 27-30,1993, pp. 151-154.
M.S. Ahmed, "Comparison of Noisy Speech Enhancement
Algorithms in Terms of LPC Perturbation," IEEE Trans, on
Acoustics, Speech, and Signal Processing, 37(1):121-125,
Primary Examiner—Allen R. MacDonald
Assistant Examiner—Susan Wieland
Attorney, Agent, or Firm—Jay H. Maioli
A noise reducing method for speech signals is provided in which the probability of speech occurring is calculated by spectral subtraction of subtracting the estimated noise spectrum from the spectrum of the input signal, and the maximum likelihood filter is adaptively controlled based upon the calculated speech occurrence probability. Adjustment to an optimum suppression factor may be achieved depending on the SNR of the input speech signal, so that is it unnecessary for the user to effect adjustment prior to practical application. In addition, a method for detecting the noise domain is provided in which the value th employed for finding the threshold value Tti1 for noise domain discrimination is calculated using the RMS value of the current frame or the value th of the previous frame multiplied by the coefficient a, whichever is smaller, and the coefficient a is changed over depending on the RMS value of the current frame. Noise domain (Hscrimination by an optimum threshold value responsive to the input signal may be achieved without producing mistaken judgement even on the occasion of noise level fluctuations.
7 Claims, 9 Drawing Sheets
SOFT DISCRIMINATION SUPPRESSION