HMM-based speech enhancement using pitch period information in voiced speech segments

Abstract In this paper a sample rate conversion algorithm which allows for continuously changing resampling ratio has been presented. The proposed implementation is based on a variable fractional delay filter which is implemented by means of a Farrow structure. Coefficients of this structure are computed on the basis of fractional delay filters which are designed using the offset window method. The proposed approach allows us to freely change the instantaneous resampling ratio during processing. Using such an algorithm we can simulate recording of audio on magnetic tape with nonuniform velocity as well as remove such distortions. We have demonstrated capabilities of the proposed approach based on the example of speech signal processing with a resampling ratio which was computed on the basis of estimated fundamental frequency of voiced speech segments.

Download Full-text

Open Quotient Measurements Based on Multiscale Product of Speech Signal Wavelet Transform

Research Letters in Signal Processing ◽

10.1155/2007/62521 ◽

2007 ◽

Vol 2007 ◽

pp. 1-5 ◽

Cited By ~ 15

Author(s):

Aïcha Bouzid ◽

Noureddine Ellouze

Keyword(s):

Wavelet Transform ◽

Relative Error ◽

Speech Processing ◽

Speech Signal ◽

Accurate Estimation ◽

Pitch Period ◽

Open Time ◽

Wide Range ◽

Voiced Speech ◽

Egg Signal

This paper describes a multiscale product method (MPM) for open quotient measure in voiced speech. The method is based on determining the glottal closing and opening instants. The proposed approach consists of making the products of wavelet transform of speech signal at different scales in order to enhance the edge detection and parameter estimation. We show that the proposed method is effective and robust for detecting speech singularity. Accurate estimation of glottal closing instants (GCIs) and opening instants (GOIs) is important in a wide range of speech processing tasks. In this paper, accurate estimation of GCIs and GOIs is used to measure the local open quotient (Oq) which is the ratio of the open time by the pitch period. Multiscale product operates automatically on speech signal; the reference electroglottogram (EGG) signal is used for performance evaluation. The ratio of good GCI detection is 95.5% and that of GOI is 76%. The pitch period relative error is 2.6% and the open phase relative error is 5.6%. The relative error measured on open quotient reaches 3% for the whole Keele database.

Download Full-text

STFT Phase Reconstruction in Voiced Speech for an Improved Single-Channel Speech Enhancement

IEEE/ACM Transactions on Audio Speech and Language Processing ◽

10.1109/taslp.2014.2354236 ◽

2014 ◽

Vol 22 (12) ◽

pp. 1931-1940 ◽

Cited By ~ 78

Author(s):

Martin Krawczyk ◽

Timo Gerkmann

Keyword(s):

Speech Enhancement ◽

Single Channel ◽

Phase Reconstruction ◽

Voiced Speech

Download Full-text

Estimating the pitch period of voiced speech

Electronics Letters ◽

10.1049/el:19800327 ◽

1980 ◽

Vol 16 (12) ◽

pp. 464 ◽

Cited By ~ 4

Author(s):

E. Ambikairajah ◽

M.J. Carey ◽

G. Tattersall

Keyword(s):

Pitch Period ◽

Voiced Speech

Download Full-text

Speech Enhancement Method under Low SNR

Advanced Materials Research ◽

10.4028/www.scientific.net/amr.457-458.1490 ◽

2012 ◽

Vol 457-458 ◽

pp. 1490-1493

Author(s):

Wen Long Cai ◽

Guang Ma

Keyword(s):

Speech Enhancement ◽

Duffing Oscillator ◽

Sinusoidal Signal ◽

Test Results ◽

Pitch Period ◽

Enhancing Effect ◽

Low Snr ◽

Harmonic Method ◽

Speech Distortion ◽

Enhancement Method

In this document, the pitch period is detected according to the sensitivity to weak sinusoidal signal and strong immunity ability to noise of Duffing oscillator, at the background of strong noise. And then enhancing pitch obtained by harmonic method. The test results show that the enhancing effect of this method is obvious under low SNR condition, and speech distortion is small.

Download Full-text

VOICED SPEECH ENHANCEMENT BASED ON ADAPTIVE FILTERING OF SELECTED INTRINSIC MODE FUNCTIONS

Advances in Adaptive Data Analysis ◽

10.1142/s1793536910000409 ◽

2010 ◽

Vol 02 (01) ◽

pp. 65-80 ◽

Cited By ~ 7

Author(s):

KAIS KHALDI ◽

MONIA TURKI-HADJ ALOUANE ◽

ABDEL-OUAHAB BOUDRAA

Keyword(s):

Speech Enhancement ◽

Additive White Gaussian Noise ◽

Main Idea ◽

Weighted Average ◽

Intrinsic Mode Functions ◽

Noisy Signal ◽

Denoising Method ◽

Low Frequencies ◽

Perceptual Evaluation ◽

Voiced Speech

In this paper a new method for voiced speech enhancement combining the Empirical Mode Decomposition (EMD) and the Adaptive Center Weighted Average (ACWA) filter is introduced. Noisy signal is decomposed adaptively into intrinsic oscillatory components called Intrinsic Mode Functions (IMFs). Since voiced speech structure is mostly distributed on both medium and low frequencies, the shorter scale IMFs of the noisy signal are beneath noise, however the longer scale ones are less noisy. Therefore, the main idea of the proposed approach is to only filter the shorter scale IMFs, and to keep the longer scale ones unchanged. In fact, the filtering of longer scale IMFs will introduce distortion rather than reducing noise. The denoising method is applied to several voiced speech signals with different noise levels and the results are compared with wavelet approach, ACWA filter and EMD–ACWA (filtering of all IMFs using ACWA filter). Relying on exhaustive simulations, we show the efficiency of the proposed method for reducing noise and its superiority over other denoising methods, i.e. to improve Signal-to-Noise Ratio (SNR), and to offer better listening quality based on a Perceptual Evaluation of Speech Quality (PESQ). The present study is limited to signals corrupted by additive white Gaussian noise.

Download Full-text