Noisy Speech Enhancement Using Harmonic-Noise Model and Codebook-Based Post-Processing

Zavarehei, Esfandiar; Vaseghi, Saeed; Yan, Qing

doi:10.1109/tasl.2007.894516

Cited by 42 publications

(29 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is noticed that Authors done the research By ignoring the phase information Wiener-like filtering [1]- [2] focus on the magnitude or a power TF, which works good but output signals with some noise as musical noise. It has been seen that frequency resolution was limited by that of the TF transform.…”

Section: Review Of Literaturementioning

confidence: 99%

See 1 more Smart Citation

Review on Multichannel HR-NMF Modeling for Non-Stationary Signals

2016

IJSR

View full text Add to dashboard Cite

Engineering in the technosevia era concerns with the high speed with clarity. It may be either in the signals or in the Images. Golden point is to achieve the accuracy and purity of output. In this paper we studied the signal processing for the part of NMF and multichannel. Along with this we found that numbers of researchers are still working with single channel signal system. The way of implementations of the researchers is different but final aim to achieve the noise free signals. In this paper we are trying to study the multichannel signals and its separation for better output without noise. Most of the researchers use filters and NMF. The observations and study of the various researchers shows that nonnegative matrix factorization (NMF) is noticeable in the spectrograms. This study will take a short review of existing system and will find out the strength and drawbacks of the studied system. This study of review paper will help to minimize the problems of existing system in the implementation phase of our new approach to words the multichannel HR -NMF modeling for Non stationary signals.

show abstract

Section: Review Of Literaturementioning

confidence: 99%

“…Esfandiar Zavarehei and et;al , [1] researchers addressed the problem of lost to noise or suppressed by noise reduction and tried to enhance the performance of conventional speech by using the harmonic noise model. They worked with amplitude, frequency, and harmonicity of sub bands of speech spectrum.…”

Section: Review Of Literaturementioning

confidence: 99%

Review on Multichannel HR-NMF Modeling for Non-Stationary Signals

2016

IJSR

View full text Add to dashboard Cite

show abstract

“…1͒. Harmonics regeneration can be implemented using adaptive comb filtering ͑Nehorai and Porat, 1986͒ techniques, nonlinear functions ͑Plapous et al, 2006͒, and codebook-based techniques that capitalize on the fact that the harmonic amplitudes are highly correlated ͑Chu, 2004; Zavarehei et al, 2007͒. Once the harmonic amplitudes are estimated, it is straightforward to estimate the partials adjacent to the main harmonics using the Gaussian model shown in Fig. 3.…”

Section: A Practical Implementationmentioning

confidence: 99%

On the importance of preserving the harmonics and neighboring partials prior to vocoder processing: Implications for cochlear implants

Loizou

2010

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

Pre-processing based noise-reduction algorithms used for cochlear implants ͑CIs͒ can sometimes introduce distortions which are carried through the vocoder stages of CI processing. While the background noise may be notably suppressed, the harmonic structure and/or spectral envelope of the signal may be distorted. The present study investigates the potential of preserving the signal's harmonic structure in voiced segments ͑e.g., vowels͒ as a means of alleviating the negative effects of pre-processing. The hypothesis tested is that preserving the harmonic structure of the signal is crucial for subsequent vocoder processing. The implications of preserving either the main harmonic components occurring at multiples of F0 or the main harmonics along with adjacent partials are investigated. This is done by first pre-processing noisy speech with a conventional noise-reduction algorithm, regenerating the harmonics, and vocoder processing the stimuli with eight channels of stimulation in steady speech-shaped noise. Results indicated that preserving the main low-frequency harmonics ͑spanning 1 or 3 kHz͒ alone was not beneficial. Preserving, however, the harmonic structure of the stimulus, i.e., the main harmonics along with the adjacent partials, was found to be critically important and provided substantial improvements ͑41 percentage points͒ in intelligibility.

show abstract

“…Instead of the time-domain signal, often low-dimensional parameter vectors are used as an intermediate representation of the sources as this leads to better and faster training of the codebooks and faster separation and enhancement algorithms. Some examples of VQ-based enhancement and separation methods are [1][2][3][4] and [5][6][7][8][9], respectively. In finding the parameters of the intermediate representation, standard estimation algorithms such as the maximum likelihood estimator based on well-known metrics can be used.…”

Section: Introductionmentioning

confidence: 99%

A new metric for VQ-based speech enhancement and separation

Christensen

Mowlaee

2011

2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Speech enhancement and separation algorithms frequently employ two-stage processing schemes, where the signal is first mapped to an intermediate low-dimensional parametric description. Then, these parameters are mapped to vectors in codebooks trained on individual noise-free sources using a vector quantizer. To obtain accurate parameters, one must employ an estimator that takes the signal characteristics into account. An open question is, however, how to derive metrics for use in the vector quantization process. In this paper, we present and derive a new metric aimed at exactly this, and we exemplify and demonstrate its use in sinusoidal modeling. The metric takes into account that parameters may have different uncertainties and dependencies associated with them and thus leads to more accurate estimates, as is demonstrated in experiments. Moreover, we incorporate the metric in a recently proposed speech separation algorithm and compare its performance to state-of-the-art methods.

show abstract

Noisy Speech Enhancement Using Harmonic-Noise Model and Codebook-Based Post-Processing

Cited by 42 publications

References 27 publications

Review on Multichannel HR-NMF Modeling for Non-Stationary Signals

Review on Multichannel HR-NMF Modeling for Non-Stationary Signals

On the importance of preserving the harmonics and neighboring partials prior to vocoder processing: Implications for cochlear implants

A new metric for VQ-based speech enhancement and separation

Contact Info

Product

Resources

About