SNR estimation based on amplitude modulation analysis with applications to noise suppression

Tchorz, Jürgen; Kollmeier, Birger

doi:10.1109/tsa.2003.811542

Cited by 82 publications

(45 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This combines amplitude modulation spectrum (AMS) [25,26,19], relative spectral transformed perceptual linear prediction (RASTA-PLP) [27] and mel-frequency cepstral coefficients (MFCCs) [28] with a gammatone filterbank (GFB) [17]. The specific implementation is taken from [18] where AMS features are computed from 32 ms frames with 10 ms frame shift to give a 15-D vector.…”

Section: Complementary Feature Set (Arpmg)mentioning

confidence: 99%

A Comparison of Perceptually Motivated Loss Functions for Binary Mask Estimation in Speech Separation

Websdale¹,

Milner²

2017

Interspeech 2017

View full text Add to dashboard Cite

This work proposes and compares perceptually motivated loss functions for deep learning based binary mask estimation for speech separation. Previous loss functions have focused on maximising classification accuracy of mask estimation but we now propose loss functions that aim to maximise the hit minus false-alarm (HIT-FA) rate which is known to correlate more closely to speech intelligibility. The baseline loss function is binary cross-entropy (CE), a standard loss function used in binary mask estimation, which maximises classification accuracy. We propose first a loss function that maximises the HIT-FA rate instead of classification accuracy. We then propose a second loss function that is a hybrid between CE and HIT-FA, providing a balance between classification accuracy and HIT-FA rate. Evaluations of the perceptually motivated loss functions with the GRID database show improvements to HIT-FA rate and ESTOI across babble and factory noises. Further tests then explore application of the perceptually motivated loss functions to a larger vocabulary dataset.

show abstract

Section: Complementary Feature Set (Arpmg)mentioning

confidence: 99%

A Comparison of Perceptually Motivated Loss Functions for Binary Mask Estimation in Speech Separation

Websdale¹,

Milner²

2017

Interspeech 2017

View full text Add to dashboard Cite

show abstract

“…(5) is employed in [45]. For many approaches, a mel filterbank [44] or similar transformations [46] are applied to the spectrum. This results in a perceptually motivated representation of the spectrogram.…”

Section: Modulationmentioning

confidence: 99%

“…The amplitude modulation spectrogram (AMS) [46] evaluates the modulation for multiple frequencies. For this, a DFT is applied to each band of the spectrogram.…”

Section: Modulationmentioning

confidence: 99%

Features for voice activity detection: a comparative analysis

Graf

Herbig²,

Buck³

et al. 2015

EURASIP J. Adv. Signal Process.

View full text Add to dashboard Cite

In many speech signal processing applications, voice activity detection (VAD) plays an essential role for separating an audio stream into time intervals that contain speech activity and time intervals where speech is absent. Many features that reflect the presence of speech were introduced in literature. However, to our knowledge, no extensive comparison has been provided yet. In this article, we therefore present a structured overview of several established VAD features that target at different properties of speech. We categorize the features with respect to properties that are exploited, such as power, harmonicity, or modulation, and evaluate the performance of some dedicated features. The importance of temporal context is discussed in relation to latency restrictions imposed by different applications. Our analyses allow for selecting promising VAD features and finding a reasonable trade-off between performance and complexity.

show abstract

“…Note that 1) the reported decision thresholds of the referenced VADs are the optimal ones over the entire test wave file; 2) the soft output of the Ying VAD is an average of the hard decisions of all sub-bands; 3) the 10 seconds' audio segment is randomly chosen from the test set. modulation spectrograms (AMS) [47], [48]. 3 , 4 The attributes of the features are listed in Table I.…”

Section: ) Acoustic Features For Vadmentioning

confidence: 99%