Single Channel Phase‐Aware Signal Processing in Speech Communication: Theory and Practice

Mowlaee, Pejman; Kulmer, Josef; Stahl, Johannes; Mayer, Florian

doi:10.1002/9781119238805

Cited by 32 publications

(28 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where 2 ½À 0 =2; 0 =2 is the integer. Equation (13) shows that the harmonic frequency f h;n and the frame shift L resolve the harmonic phase, and the non-harmonic phase is calculated from the harmonic phase through the phase property of the window W , as shown in Fig. 4.…”

Section: Methods Using a Temporal Constraint Betweenmentioning

confidence: 99%

“…3 consider only harmonics up to 4 kHz, that is, the number of harmonics H ¼ b4;000= f 0;n c, and therefore it is valid to use the PESQ to evaluate them. In addition, short-time objective intelligibility (STOI) [22] and unwrapped phase RMSE (UnRMSE) [13] have been used for evaluating quality. The STOI is difficult to discern during the performance of estimation accuracy unless the estimation accuracy is greatly improved.…”

Section: Perceived Quality Measurementioning

confidence: 99%

“…Advances in deep learning have accelerated phase reconstruction for speech synthesis with deep neural networks (DNNs) [11] and the time-frequency mask estimation using phase information with DNNs [12]. A textbook on phase processing was also published recently [13].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Speech enhancement using harmonic-structure-based phase reconstruction

Wakabayashi

2019

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

Recent work has shown that phase information is useful for further improving the performance of speech enhancement, source separation, and speech synthesis. In the speech enhancement field, the combination of amplitude and phase estimations improves the perceived quality more than only amplitude estimation. In this paper, we review two harmonic-structure-based phase estimation methods with temporal and frequency constraints on the harmonic speech phase. In addition, we describe important parameters for phase estimation, such as the frame shift length and window function of the short-time Fourier transform. Subjective experiments using listening tests and future work for phase processing are briefly described.

show abstract

Section: Methods Using a Temporal Constraint Betweenmentioning

confidence: 99%

Section: Perceived Quality Measurementioning

confidence: 99%

See 1 more Smart Citation

Speech enhancement using harmonic-structure-based phase reconstruction

Wakabayashi

2019

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

show abstract

“…Several early studies point out the unimportance [1] and importance [2][3][4] of spectral phase information from a perceptual viewpoint. Several more recent studies reported positive impact of spectral phase in different speech processing applications including speech enhancement [5][6][7][8], speech intelligibility prediction [9,10] and ASR [11] (see [12,13] for an overview). Conventional automatic speech recognition (ASR) systems often rely solely on the magnitude spectrum and are built upon short-time amplitude-derived features [14].…”

Section: Introductionmentioning

confidence: 99%

Phase-Aware Signal Processing for Automatic Speech Recognition

et al. 2016

Self Cite

View full text Add to dashboard Cite

Conventional automatic speech recognition (ASR) often neglects the spectral phase information in its front-end and feature extraction stages. The aim of this paper is to show the impact that enhancement of the noisy spectral phase has on ASR accuracy when dealing with speech signals corrupted with additive noise. Apart from proof-of-concept experiments using clean spectral phase, we also present a phase enhancement method as a phase-aware front-end and modified group delay as a phaseaware feature extractor, and the combination thereof. In experiments, we demonstrate the improved performance for each individual component and their combination, compared to the conventional phase-unaware Mel Frequency Cepstral Coefficients (MFCCs)-based ASR. We observe that the estimated phase information used in the front-end or feature extraction component improves the ASR word accuracy rate (WAR) by 20.98 % absolute for noise corrupted speech (averaged over SNRs ranging from 0 to 20 dB).

show abstract

“…Although phase information was eliminated in the present envelope-based speech synthesis, many recent studies found that phase (or temporal fine-structure waveform, which contains phase fluctuation information) cue carries important information for tone identification, speech perception in noise, and music appreciation [e.g., [8][9][10][11][12]. Unfortunately, our knowledge on the effect of phase information to speech perception is still limited.…”

Section: Introductionmentioning

confidence: 99%

Comparing the Contributions of Amplitude and Phase to Speech Intelligibility in a Vocoder-Based Speech Synthesis Model

Chen

Chiao²

2016

Interspeech 2016

View full text Add to dashboard Cite

Vocoder-based speech synthesis model has been long used to assess the contribution of acoustic cue for speech recognition. This study compared the perceptual contributions of amplitude and phase by using two types of stimuli, i.e., amplitude-and phase-based vocoded stimuli. The amplitude-based vocoded stimuli were synthesized by preserving amplitude fluctuation cue but discarding phase cue (i.e., setting phase to zero), while the phase-based vocoded stimuli were synthesized by preserving phase cue and discarding amplitude cue (i.e., setting amplitude to unit). Listening experiments with normalhearing participants showed consistent findings with earlier studies that the intelligibility scores of both amplitude-and phase-based vocoded stimuli increased when using a large number of channels in vocoder-based speech synthesis. In addition, at all tested conditions, the intelligibility scores of amplitude-based vocoded stimuli were significantly larger than those of phase-based vocoded stimuli, suggesting that amplitude might carry more perceptual contribution than phase. This intelligibility advantage of amplitude over phase may be attributed to the difference in the amount of envelope information contained in the two types of vocoded stimuli.

show abstract

Single Channel Phase‐Aware Signal Processing in Speech Communication: Theory and Practice

Cited by 32 publications

References 0 publications

Speech enhancement using harmonic-structure-based phase reconstruction

Speech enhancement using harmonic-structure-based phase reconstruction

Phase-Aware Signal Processing for Automatic Speech Recognition

Comparing the Contributions of Amplitude and Phase to Speech Intelligibility in a Vocoder-Based Speech Synthesis Model

Contact Info

Product

Resources

About