INTERSPEECH 2021 Deep Noise Suppression Challenge

Reddy, Chandan K.; Dubey, Harishchandra; Koishida, Kazuhito; Nair, Arun Asokan; Gopal, Vishak; Cutler, Ross; Braun, Sebastian; Gamper, Hannes; Aichner, Robert; Srinivasan, Sriram

doi:10.21437/interspeech.2021-1609

Cited by 116 publications

(56 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The WSJ0 training speech utterances were mixed with a diverse set of noises coming from the FreeField72k dataset, which was also used in [ 25 ]. Motivated by the popularity of the freefield1010 dataset [ 31 ], which was also included in the DNS challenge [ 32 ], the FreeField72k extends its predecessor by taking a much bigger sample of the free field recordings from the source database. In the process of creation of the FreeField72k, the Freesound online database [ 33 ] was queried, and all of the recordings tagged with the “field-recording” tag were downloaded.…”

Section: Methodsmentioning

confidence: 99%

Speech Enhancement by Multiple Propagation through the Same Neural Network

Grzywalski

Drgas

2022

Sensors

View full text Add to dashboard Cite

Monaural speech enhancement aims to remove background noise from an audio recording containing speech in order to improve its clarity and intelligibility. Currently, the most successful solutions for speech enhancement use deep neural networks. In a typical setting, such neural networks process the noisy input signal once and produces a single enhanced signal. However, it was recently shown that a U-Net-based network can be trained in such a way that allows it to process the same input signal multiple times in order to enhance the speech even further. Unfortunately, this was tested only for two-iteration enhancement. In the current research, we extend previous efforts and demonstrate how the multi-forward-pass speech enhancement can be successfully applied to other architectures, namely the ResBLSTM and Transformer-Net. Moreover, we test the three architectures with up to five iterations, thus identifying the method’s limit in terms of performance gain. In our experiments, we used the audio samples from the WSJ0, Noisex-92, and DCASE datasets and measured speech enhancement quality using SI-SDR, STOI, and PESQ. The results show that performing speech enhancement up to five times still brings improvements to speech intelligibility, but the gain becomes smaller with each iteration. Nevertheless, performing five iterations instead of two gives additional a 0.6 dB SI-SDR and four-percentage-point STOI gain. However, these increments are not equal between different architectures, and the U-Net and Transformer-Net benefit more from multi-forward pass compared to ResBLSTM.

show abstract

Section: Methodsmentioning

confidence: 99%

Speech Enhancement by Multiple Propagation through the Same Neural Network

Grzywalski

Drgas

2022

Sensors

View full text Add to dashboard Cite

show abstract

“…We first take a comprehensive ablation study on the proposed model on the DNS-2020 dataset [1]. Then our model is trained, integrated with the post-processing module and evaluated with the Interspeech 2021 DNS challenge dataset (DNS-2021) [2] to show its performance on more complicated and real acoustic scenarios. We also compare other competitive models (such as PercepNet [27]) with our model on Voice Bank + DEMAND dataset [28] as these models have PESQ scores reported on this set.…”

Section: Datasetsmentioning

confidence: 99%

“…Such data-driven approaches have become the mainstream because of their strong noise reduction abilities (especially for non-stationary noise) learned from simulated cleannoisy speech pairs. The recent deep noise suppression challenge (DNS) series [1,2] have benchmarked many state-of-the-art DLbased speech enhancers, especially for real-time ones for speech communications, through subjective listening test and promising performance has been reported.…”

Section: Introductionmentioning

confidence: 99%

“…tion) is how to preserve perceived subjective speech quality to the best extent while greatly suppressing the noise interference. Some recent studies, including the new DNS challenge initiative [2], have pointed out that many neural noise suppressors are very good at suppressing noise, but do not improve the quality of speech, even introducing apparent speech distortions. In this paper, we study the problem by revising our recently proposed deep complex convolution recurrent network (DCCRN) [3].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DCCRN+: Channel-Wise Subband DCCRN with SNR Estimation for Speech Enhancement

Lv¹,

Zhang

et al. 2021

Interspeech 2021

View full text Add to dashboard Cite

Deep complex convolution recurrent network (DCCRN), which extends CRN with complex structure, has achieved superior performance in MOS evaluation in Interspeech 2020 deep noise suppression challenge (DNS2020). This paper further extends DCCRN with the following significant revisions. We first extend the model to sub-band processing where the bands are split and merged by learnable neural network filters instead of engineered FIR filters, leading to a faster noise suppressor trained in an end-to-end manner. Then the LSTM is further substituted with a complex TF-LSTM to better model temporal dependencies along both time and frequency axes. Moreover, instead of simply concatenating the output of each encoder layer to the input of the corresponding decoder layer, we use convolution blocks to first aggregate essential information from the encoder output before feeding it to the decoder layers. We specifically formulate the decoder with an extra a priori SNR estimation module to maintain good speech quality while removing noise. Finally a post-processing module is adopted to further suppress the unnatural residual noise. The new model, named DCCRN+, has surpassed the original DCCRN as well as several competitive models in terms of PESQ and DNSMOS, and has achieved superior performance in the new Interspeech 2021 DNS challenge.

show abstract

“…However, the background noise and reverberation degrade the speech and transcription quality. As these acoustic distortions can hamper communications and thus productivity, the speech enhancement (SE) field has drawn a lot of renewed attention recently [1,2].…”

Section: Introductionmentioning

confidence: 99%

Fast Real-time Personalized Speech Enhancement: End-to-End Enhancement Network (E3Net) and Knowledge Distillation

Thakker¹,

Eskimez²,

Yoshioka³

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper investigates how to improve the runtime speed of personalized speech enhancement (PSE) networks while maintaining the model quality. Our approach includes two aspects: architecture and knowledge distillation (KD). We propose an end-to-end enhancement (E3Net) model architecture, which is 3× faster than a baseline STFT-based model. Besides, we use KD techniques to develop compressed student models without significantly degrading quality. In addition, we investigate using noisy data without reference clean signals for training the student models, where we combine KD with multi-task learning (MTL) using an automatic speech recognition (ASR) loss. Our results show that E3Net provides better speech and transcription quality with a lower target speaker over-suppression (TSOS) rate than the baseline model. Furthermore, we show that the KD methods can yield student models that are 2 − 4× faster than the teacher and provides reasonable quality. Combining KD and MTL improves the ASR and TSOS metrics without degrading the speech quality.

show abstract

INTERSPEECH 2021 Deep Noise Suppression Challenge

Cited by 116 publications

References 12 publications

Speech Enhancement by Multiple Propagation through the Same Neural Network

Speech Enhancement by Multiple Propagation through the Same Neural Network

DCCRN+: Channel-Wise Subband DCCRN with SNR Estimation for Speech Enhancement

Fast Real-time Personalized Speech Enhancement: End-to-End Enhancement Network (E3Net) and Knowledge Distillation

Contact Info

Product

Resources

About