A Novel Approach to Speech Enhancement Based on Deep Neural Networks
Minimum mean-square error (MMSE) approaches have been shown to achieve state-of-the-art performance on the task of speech enhancement. However, MMSE approaches lack the ability to accurately estimate non-stationary noise sources. In this paper, a long short-term memory fully convolutional network...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Stefan cel Mare University of Suceava
2022-05-01
|
| Series: | Advances in Electrical and Computer Engineering |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.4316/AECE.2022.02009 |
| _version_ | 1818231864017551360 |
|---|---|
| author | SALEHI, M. MIRZAKUCHAKI, S. |
| author_facet | SALEHI, M. MIRZAKUCHAKI, S. |
| author_sort | SALEHI, M. |
| collection | DOAJ |
| description | Minimum mean-square error (MMSE) approaches have been shown to achieve state-of-the-art performance on the
task of speech enhancement. However, MMSE approaches lack the ability to accurately estimate non-stationary
noise sources. In this paper, a long short-term memory fully convolutional network (LSTM-FCN) is utilized
to accurately estimate a priori signal-to-noise ratio (SNR) since the speech enhancement performance of
an MMSE approach improves with the accuracy of the used a priori SNR estimator. The proposed MMSE approach
makes no assumptions about the characteristics of the noise or the speech. MMSE approaches that utilize
the LSTM-FCN estimator are evaluated using the mean opinion score of the perceptual evaluation of speech
quality (PESQ) and the short-time objective intelligibility (STOI) measures of speech. The experimental
investigation shows that the speech enhancement performance of an MMSE approach that utilizes LSTM-FCN
estimator significantly increases. |
| first_indexed | 2024-12-12T10:57:09Z |
| format | Article |
| id | doaj.art-80d219a923d74590a31d22c42721ed99 |
| institution | Directory Open Access Journal |
| issn | 1582-7445 1844-7600 |
| language | English |
| last_indexed | 2024-12-12T10:57:09Z |
| publishDate | 2022-05-01 |
| publisher | Stefan cel Mare University of Suceava |
| record_format | Article |
| series | Advances in Electrical and Computer Engineering |
| spelling | doaj.art-80d219a923d74590a31d22c42721ed992022-12-22T00:26:38ZengStefan cel Mare University of SuceavaAdvances in Electrical and Computer Engineering1582-74451844-76002022-05-01222717810.4316/AECE.2022.02009A Novel Approach to Speech Enhancement Based on Deep Neural NetworksSALEHI, M.MIRZAKUCHAKI, S.Minimum mean-square error (MMSE) approaches have been shown to achieve state-of-the-art performance on the task of speech enhancement. However, MMSE approaches lack the ability to accurately estimate non-stationary noise sources. In this paper, a long short-term memory fully convolutional network (LSTM-FCN) is utilized to accurately estimate a priori signal-to-noise ratio (SNR) since the speech enhancement performance of an MMSE approach improves with the accuracy of the used a priori SNR estimator. The proposed MMSE approach makes no assumptions about the characteristics of the noise or the speech. MMSE approaches that utilize the LSTM-FCN estimator are evaluated using the mean opinion score of the perceptual evaluation of speech quality (PESQ) and the short-time objective intelligibility (STOI) measures of speech. The experimental investigation shows that the speech enhancement performance of an MMSE approach that utilizes LSTM-FCN estimator significantly increases.http://dx.doi.org/10.4316/AECE.2022.02009long short-term memorymachine learningmean square error methodsrecurrent neural networksspeech enhancement |
| spellingShingle | SALEHI, M. MIRZAKUCHAKI, S. A Novel Approach to Speech Enhancement Based on Deep Neural Networks Advances in Electrical and Computer Engineering long short-term memory machine learning mean square error methods recurrent neural networks speech enhancement |
| title | A Novel Approach to Speech Enhancement Based on Deep Neural Networks |
| title_full | A Novel Approach to Speech Enhancement Based on Deep Neural Networks |
| title_fullStr | A Novel Approach to Speech Enhancement Based on Deep Neural Networks |
| title_full_unstemmed | A Novel Approach to Speech Enhancement Based on Deep Neural Networks |
| title_short | A Novel Approach to Speech Enhancement Based on Deep Neural Networks |
| title_sort | novel approach to speech enhancement based on deep neural networks |
| topic | long short-term memory machine learning mean square error methods recurrent neural networks speech enhancement |
| url | http://dx.doi.org/10.4316/AECE.2022.02009 |
| work_keys_str_mv | AT salehim anovelapproachtospeechenhancementbasedondeepneuralnetworks AT mirzakuchakis anovelapproachtospeechenhancementbasedondeepneuralnetworks AT salehim novelapproachtospeechenhancementbasedondeepneuralnetworks AT mirzakuchakis novelapproachtospeechenhancementbasedondeepneuralnetworks |