Analysis of DNN Speech Signal Enhancement for Robust Speaker Recognition

Novotny, Ondrej; Plchot, Oldřich; Glembek, Ondřej; Černocký, Jaň; Burget, Lukáš

doi:10.1016/j.csl.2019.06.004

Cited by 43 publications

(22 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this paper, DCF is the average of two minimum DCF scores when Ptarget, a priori probability of the specified target speaker, is 0.01 and 0.001. Performance comparison is done with DAE-based speech enhancement [8,10,9]. We use an 8-layer time-delay neural network (TDNN) [16] with 1000 hidden units per layer for enhancement.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

VoiceID Loss: Speech Enhancement for Speaker Verification

2019

View full text Add to dashboard Cite

In this paper, we propose VoiceID loss, a novel loss function for training a speech enhancement model to improve the robustness of speaker verification. In contrast to the commonly used loss functions for speech enhancement such as the L2 loss, the VoiceID loss is based on the feedback from a speaker verification model to generate a ratio mask. The generated ratio mask is multiplied pointwise with the original spectrogram to filter out unnecessary components for speaker verification. In the experiments, we observed that the enhancement network, after training with the VoiceID loss, is able to ignore a substantial amount of time-frequency bins, such as those dominated by noise, for verification. The resulting model consistently improves the speaker verification system on both clean and noisy conditions.

show abstract

Section: Methodsmentioning

confidence: 99%

“…For these reasons, only a few studies have explored speech enhancement for speaker verification [7,6] and most recent studies are based on the i-vector approach [8,9,10]. They used a Denoising Autoencoder (DAE) to generate an enhanced signal from the noisy signal.…”

Section: Introductionmentioning

confidence: 99%

VoiceID Loss: Speech Enhancement for Speaker Verification

2019

View full text Add to dashboard Cite

show abstract

“…The proposed LSTM with 20% to 30% relative improvement of EER, has given better results in comparison to other methods used in this research. In [6] a denoising autoencoder is used for joint compensation of additive noise and reverberation in the xvector framework. In this research, the DAE reconstructs the clean version of the magnitude spectrum.…”

Section: Related Workmentioning

confidence: 99%

“…Approaches based on speech signal enhancement [6] and denoising techniques at the speaker modeling level are often proposed to reduce the impact of noise and reverberation on speaker recognition systems [7].…”

Section: Introductionmentioning

confidence: 99%

Compensate multiple distortions for speaker recognition systems

Mohammadamini

Matrouf

Bonastre

et al. 2021

2021 29th European Signal Processing Conference (EUSIPCO)

View full text Add to dashboard Cite

The performance of speaker recognition systems reduces dramatically in severe conditions in the presence of additive noise and/or reverberation. In some cases, there is only one kind of domain mismatch like additive noise or reverberation, but in many cases, there are more than one distortion. Finding a solution for domain adaptation in the presence of different distortions is a challenge. In this paper we investigate the situation in which there is none, one or more of the following distortions: early reverberation, full reverberation, additive noise. We propose two configurations to compensate for these distortions. In the first one a specific denoising autoencoder is used for each distortion. In the second configuration, a denoising autoencoder is used to compensate for all of these distortions simultaneously. Our experiments show that, in the co-existence of noise and reverberation, the second configuration gives better results. For example, with the second configuration we obtained 76.6% relative improvement of EER for utterances longer than 12 seconds. For other situations in the presence of only one distortion, the second configuration gives almost the same results achieved by using a specific model for each distortion.

show abstract

“…Denoising autoencoder is another DNN method used as a preprocessing step in speaker recognition. In [10] a deep feedforward autoencoder trained in order to transform the noisy log magnitude spectrum to a clean compact encoded signal in the same domain. In another paper, a DNN architecture that uses the feedback from the speaker verification (SV) system to generate a ratio mask was designed.…”

Section: Related Workmentioning

confidence: 99%

Denoising x-vectors for Robust Speaker Recognition

Mohammadamini

Matrouf

Noé

2020

The Speaker and Language Recognition Workshop (Odyssey 2020)

View full text Add to dashboard Cite

Using deep learning methods has led to significant improvement in speaker recognition systems. Introducing xvectors as a speaker modeling method has made these systems more robust. Since, in challenging environments with noise and reverberation, the performance of x-vectors systems degrades significantly, the demand for denoising techniques remains as before. In this paper, for the first time, we try to denoise the xvectors speaker embedding. Our focus is on additive noise. Firstly, we use the i-MAP method which considers that both noise and clean x-vectors have a Gaussian distribution. Then, leveraging denoising autoencoders (DAE) we try to reconstruct the clean x-vector from the corrupted version. After that, we propose two hybrid systems composed of statistical i-MAP and DAE. Finally, we propose a novel DAE architecture, named Deep Stacked DAE, composed of several DAEs where each DAE receives as input the output of its predecessor DAE concatenated with the difference between noisy x-vectors and its predecessor's output. The experiments on Fabiol corpus show that the results given by the hybrid DAE i-MAP method in several cases outperforms the conventional DAE and i-MAP methods. Also, the results for Deep Stacked DAE in most cases is better than the other proposed methods. For utterances longer than 12 seconds we achieved a 51% improvement in terms of EER with Deep Stacked DAE, and for utterances shorter than 2 seconds, Deep Stacked DAE gives 18% improvements compared to the baseline system.

show abstract

Analysis of DNN Speech Signal Enhancement for Robust Speaker Recognition

Cited by 43 publications

References 36 publications

VoiceID Loss: Speech Enhancement for Speaker Verification

VoiceID Loss: Speech Enhancement for Speaker Verification

Compensate multiple distortions for speaker recognition systems

Denoising x-vectors for Robust Speaker Recognition

Contact Info

Product

Resources

About