On the relative importance of various components of the modulation spectrum for automatic speech recognition

Kanedera, Noboru; Arai, Takayuki; Heřmanský, Hynek; Pavel, Misha

doi:10.1016/s0167-6393(99)00002-3

Cited by 120 publications

(91 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It is a bandpass filter that removes the very low frequency and high frequency components of feature trajectories. The design agrees with research findings that speech modulation frequency of 1-16Hz is most important for both human and automatic speech recognition [62][63][64][65][66][67][68]. RASTA and CMN are both able to reduce channel distortions, and they can be used in concatenation to produce better results.…”

Section: Temporal Filteringsupporting

confidence: 86%

Features and Model Adaptation Techniques for Robust Speech Recognition: A Review

Legoh¹,

Bhattacharjee²,

Tuithung³

2015

CAE

View full text Add to dashboard Cite

In this paper, major speech features used in state-of-the-art technology in speech recognition research are reviewed. Also a brief review of major technological advancements during last few decades and a trend towards development of robust speech recognition system in terms of feature and model adaptation techniques is given. It has been the dream of researchers to develop a machine that recognizes speech and understands natural language like human but the reality is that the performance of the speech recognition system drastically degrades due to various adverse conditions like noise, variability in speaker, channel, device and mismatches in training and testing. This paper may be useful as a tutorial and review on state-of-the-art techniques for feature selection, feature normalization and model adaptation techniques for development of robust speech recognition system.

show abstract

Section: Temporal Filteringsupporting

confidence: 86%

Features and Model Adaptation Techniques for Robust Speech Recognition: A Review

Legoh¹,

Bhattacharjee²,

Tuithung³

2015

CAE

View full text Add to dashboard Cite

show abstract

“…In this paper, we use an LPF cut-off frequency of 20 Hz in both equations because an important modulation region for speech perception [15] and speech recognition is from 1 to 16 Hz [16,17].…”

Section: Extraction Of the Power Envelopementioning

confidence: 99%

An improved method based on the MTF concept for restoring the power envelope from a reverberant signal

Unoki

Furukawa

Sakata

et al. 2004

Acoust. Sci. & Tech.

View full text Add to dashboard Cite

A basic method for restoring the power envelope from a reverberant signal was proposed by Hirobayashi et al. This method is based on the concept of the modulation transfer function (MTF) and does not require that the impulse response of an environment be measured. However this basic method has the following problems: (i) how to precisely extract the power envelope from the observed signal; (ii) how to determine the parameters of the impulse response of the room acoustics; and (iii) a lack of consideration as to whether the MTF concept can be applied to a more realistic signal. This paper improves this basic method with regard to these problems in order to extend this method as a first step towards the development for speech applications. We have carried out 1,500 simulations for restoring the power envelope from reverberant signals in which the power envelopes are three types of sinusoidal, harmonics, and band-limited noise and the carriers are white noise, to evaluate our improved method with regard to (i) and (ii). We then have carried out the same simulations in which the carriers are two types of carrier of white noise or harmonics with regard to (iii). Our results have shown that the improved method can adequately restore the power envelope from a reverberant signal and will be able to be applied for speech envelope restoration.

show abstract

“…This fact is further confirmed by automatic speech recognition (ASR) experiments [3] which shows that the low modulation frequency bands (0-1Hz) and high modulation frequency bands are harmful (or useless) for ASR. The most widely used cepstral-based features in state-of-art ASR system, however, are typically sampled at a rate of 100Hz, giving a 50Hz bandwidth for representing modulation energy, something that is overkill.…”

Section: Introductionmentioning

confidence: 71%

Polyphase speech recognition

Lin

Bilmes

2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

We propose a model for speech recognition that consists of multiple semi-synchronized recognizers operating on a polyphase decomposition of standard speech features. Specifically, we consider multiple out-of-phase downsampled speech features as separate streams which are modeled separately at the lowest level, and are then integrated at the higher level (words) during first-pass decoding. Our model lessens the severity of the oversampling problem in many speech recognition systems -i.e., that speech modulation energy is most important below 25Hz but a 100Hz frame rate gives a modulation bandwidth of 50Hz. Our polyphase approach moreover captures wider and more diverse dynamics within the speech signal. Our integrative network is high-level, namely it couples together and decodes word strings from different recognizers simultaneously and asynchronously. We provide preliminary results on the 10-word vocabulary version of the SVitchboard (small-vocabulary switchboard) task and show that our polyphase recognition system significantly outperforms an optimized baseline (HMM) approach.

show abstract

On the relative importance of various components of the modulation spectrum for automatic speech recognition

Cited by 120 publications

References 10 publications

Features and Model Adaptation Techniques for Robust Speech Recognition: A Review

Features and Model Adaptation Techniques for Robust Speech Recognition: A Review

An improved method based on the MTF concept for restoring the power envelope from a reverberant signal

Polyphase speech recognition

Contact Info

Product

Resources

About