Abstract
The recent development of speech enhancement methods has incorporated attention mechanisms for learning long-term speech signal dependencies. The utilization of deep convolution networks (DCN) equipped with the self-attention (SA) and transformers has showed promising results in speech enhancement (SE). While self-attention networks excel in extracting significant long-sequence contextual information in mining tasks, they may not effectively concentrate on subtle aspects within speech signals. These subtle details include temporal or spectral continuity, spectral structure, and timbre. To tackle this problem, in the current work, we propose a novel speech enhancement model based on adaptive attention. The proposed model incorporates both local and global attention modules in between a convolutional encoder and a convolutional decoder. The local attention module (LAM) integrates channel and spatial attentions, which can make the model pay more attention to the local details in the speech block, specifically the frame-level features. And the features at utterance-level are explored through a self-attention mechanism in global attention module (GAM). Different from existing transformers, the feed forward network of GAM is improved by introducing a 1D-Conv layer and Bi-directional long short-term memory (Bi-LSTM) for extracting global features, so that the network can more effectively model long sequence context. Moreover, a CNN module is also added to global attention module so that short-term noise can be reduced more effectively, based on the ability of CNN to extract local information. The proposed model stands apart from the current speech enhancement techniques that solely rely on self-attention networks. Instead, our approach models the speech signal using two different attention networks simultaneously, both local detail information and global contextual information of speech are considered, thus better extracting useful information from the speech signal. The effectiveness of the proposed model is assessed using both objective (PESQ and STOI) and subjective tests (signal distortion (CSIG), background distortion (CBAK) and overall quality (COVL)) on two distinct datasets: Voice Bank-Demand dataset and LibriSpeech dataset. The experimental findings demonstrate that our model outperformed the competing baselines on both the datasets.
Similar content being viewed by others
References
Abdulatif S, Cao R, Yang B (2022) Cmgan: Conformer-based metric-gan for monaural speech enhancement. arXiv preprint arXiv:2209.11112
Abdulbaqi J, Gu Y, Chen S et al (2020) Residual recurrent neural network for speech enhancement. ICASSP 2020–2020 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6659–6663
Abgeena A, Garg S (2023) S-lstm-att: a hybrid deep learning approach with optimized features for emotion recognition in electroencephalogram. Health inf sci syst 11(1):40
Bastanfard A, Abbasian A (2023) Speech emotion recognition in persian based on stacked autoencoder by comparing local and global features. Multimed Tools Appl. pp 1–18
Braun S, Gamper H, Reddy CK et al (2021) Towards efficient models for real-time deep noise suppression. ICASSP 2021–2021 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 656–660
Defossez A, Synnaeve G, Adi Y (2020) Real time speech enhancement in the waveform domain. arXiv preprint arXiv:2006.12847
Fu SW, Hu Ty, Tsao Y, et al (2017) Complex spectrogram enhancement by convolutional neural network with multi-metrics learning. In: 2017 IEEE 27th international workshop on machine learning for signal processing (MLSP), IEEE, pp 1–6
Fu SW, Liao CF, Tsao Y, et al (2019) Metricgan: Generative adversarial networks based black-box metric scores optimization for speech enhancement. In: International Conference on Machine Learning, PMLR, pp 2031–2041
Giri R, Isik U, Krishnaswamy A (2019) Attention wave-u-net for speech enhancement. In: 2019 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), IEEE, pp 249–253
Gnanamanickam J, Natarajan Y, KR SP (2021) A hybrid speech enhancement algorithm for voice assistance application. Sensors 21(21):7025
Han JY, Zheng WZ, Huang RJ, et al (2018) Hearing aids app design based on deep learning technology. In: 2018 11th International Symposium on Chinese Spoken Language Processing (ISCSLP), IEEE, pp 495–496
Han X, Pan M, Li Z, et al (2022) Vhf speech enhancement based on transformer. ’IEEE Trans Intell Transp Syst 3:146–152
Hao K (2020) Multimedia english teaching analysis based on deep learning speech enhancement algorithm and robust expression positioning. J Intell. Fuzzy Syst 39(2):1779–1791
He B, Wang K, Zhu WP (2022) Dbaunet: Dual-branch attention u-net for time-domain speech enhancement. In: TENCON 2022-2022 IEEE Region 10 Conference (TENCON), IEEE, pp 1–6
Hsieh TA, Wang HM, Lu X et al (2020) Wavecrn: An efficient convolutional recurrent neural network for end-to-end speech enhancement. IEEE Signal Process Lett 27:2149–2153
Hu Y (2007) Subjective evaluation and comparison of speech enhancement algorithms. Speech Commun 49:588–601
Hu Y, Loizou PC (2007) Evaluation of objective quality measures for speech enhancement. IEEE Trans Audio Speech Lang Process 16(1):229–238
Hu Y, Liu Y, Lv S, et al (2020) Dccrn: Deep complex convolution recurrent network for phase-aware speech enhancement. arXiv preprint arXiv:2008.00264
Jannu C, Vanambathina SD (????) Dct based densely connected convolutional gru for real-time speech enhancement. J Intell Fuzzy Syst (Preprint):1–14
Jannu C, Vanambathina SD (2023) An attention based densely connected u-net with convolutional gru for speech enhancement. In: 2023 3rd International conference on Artificial Intelligence and Signal Processing (AISP), IEEE, pp 1–5
Jannu C, Vanambathina SD (2023) Multi-stage progressive learning-based speech enhancement using time–frequency attentive squeezed temporal convolutional networks. Circuits, Systems, and Signal Processing pp 1–27
Jannu C, Vanambathina SD (2023) Shuffle attention u-net for speech enhancement in time domain. Int J Image Graph p 2450043
Karthik A, MazherIqbal J (2021) Efficient speech enhancement using recurrent convolution encoder and decoder. Wirel Pers Commun 119(3):1959–1973
Kim E, Seo H (2021) Se-conformer: Time-domain speech enhancement using conformer. In: Interspeech, pp 2736–2740
Kingma DP, Ba J (2014) Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980
Kishore V, Tiwari N, Paramasivam P (2020) Improved speech enhancement using tcn with multiple encoder-decoder layers. In: Interspeech, pp 4531–4535
Koizumi Y, Yatabe K, Delcroix M et al (2020) Speech enhancement using self-adaptation and multi-head self-attention. ICASSP 2020–2020 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 181–185
Kong Z, Ping W, Dantrey A et al (2022) Speech denoising in the waveform domain with self-attention. ICASSP 2022–2022 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 7867–7871
Lalitha V, Prema P, Mathew L (2010) A kepstrum based approach for enhancement of dysarthric speech. In: 2010 3rd International Congress on Image and Signal Processing, IEEE, pp 3474–3478
Li A, Yuan M, Zheng C et al (2020) Speech enhancement using progressive learning-based convolutional recurrent neural network. Appl Acoust 166:107347
Li A, Liu W, Zheng C et al (2021) Two heads are better than one: A two-stage complex spectral mapping approach for monaural speech enhancement. IEEE/ACM Trans. Audio Speech Lang Process 29:1829–1843
Li A, Zheng C, Peng R, et al (2021) On the importance of power compression and phase estimation in monaural speech dereverberation. JASA express letters 1(1)
Lin J, Van Wijngaarden AJ, Smith MC, et al (2021) Speaker-aware speech enhancement with self-attention. In: 2021 29th European Signal Processing Conference (EUSIPCO), IEEE, pp 486–490
Lin J, van Wijngaarden AJdL, Wang KC et al (2021) Speech enhancement using multi-stage self-attentive temporal convolutional networks. IEEE/ACM Trans. Audio Speech Lang Process 29:3440–3450
Macartney C, Weyde T (2018) Improved speech enhancement with the wave-u-net. arXiv preprint arXiv:1811.11307
Mehrish A, Majumder N, Bharadwaj R, et al (2023) A review of deep learning techniques for speech processing. Inf Fusion p 101869
Nossier SA, Wall J, Moniri M, et al (2020) Mapping and masking targets comparison using different deep learning based speech enhancement architectures. In: 2020 International Joint Conference on Neural Networks (IJCNN), IEEE, pp 1–8
Panayotov V, Chen G, Povey D, et al (2015) Librispeech: an asr corpus based on public domain audio books. In: 2015 IEEE international conference on acoustics, speech and signal processing (ICASSP), IEEE, pp 5206–5210
Pandey A, Wang D (2019) Tcnn: Temporal convolutional neural network for real-time speech enhancement in the time domain. ICASSP 2019–2019 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6875–6879
Pandey A, Wang D (2020) Densely connected neural network with dilated convolutions for real-time speech enhancement in the time domain. ICASSP 2020–2020 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6629–6633
Pandey A, Wang D (2021) Dense cnn with self-attention for time-domain speech enhancement. IEEE/ACM Trans Audio Speech Lang Process 29:1270–1279
Pascual S, Bonafonte A, Serra J (2017) Segan: Speech enhancement generative adversarial network. arXiv preprint arXiv:1703.09452
Phan H, McLoughlin IV, Pham L et al (2020) Improving gans for speech enhancement. IEEE Signal Process Lett 27:1700–1704
Phan H, Le Nguyen H, Chén OY et al (2021) Self-attention generative adversarial network for speech enhancement. ICASSP 2021–2021 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 7103–7107
Recommendation I (2003) Subjective test methodology for evaluating speech communication systems that include noise suppression algorithm. ITU-T recommendation p 835
Reddy CK, Dubey H, Gopal V et al (2021) Icassp 2021 deep noise suppression challenge. ICASSP 2021–2021 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6623–6627
Rethage D, Pons J, Serra X (2018) A wavenet for speech denoising. 2018 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 5069–5073
Rix AW, Beerends JG, Hollier MP, et al (2001) Perceptual evaluation of speech quality (pesq)-a new method for speech quality assessment of telephone networks and codecs. In: 2001 IEEE international conference on acoustics, speech, and signal processing. Proceedings (Cat. No. 01CH37221), IEEE, pp 749–752
Roy SK, Paliwal KK (2020) Causal convolutional encoder decoder-based augmented kalman filter for speech enhancement. In: 2020 14th International Conference on Signal Processing and Communication Systems (ICSPCS), IEEE, pp 1–7
Shahnawazuddin S, Deepak K, Pradhan G et al (2017) Enhancing noise and pitch robustness of children’s asr. 2017 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 5225–5229
Soni MH, Shah N, Patil HA (2018) Time-frequency masking-based speech enhancement using generative adversarial network. In: 2018 IEEE international conference on acoustics, speech and signal processing (ICASSP), IEEE, pp 5039–5043
Strake M, Defraene B, Fluyt K et al (2020) Fully convolutional recurrent networks for speech enhancement. ICASSP 2020–2020 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6674–6678
Taal CH, Hendriks RC, Heusdens R, et al (2010) A short-time objective intelligibility measure for time-frequency weighted noisy speech. In: 2010 IEEE international conference on acoustics, speech and signal processing, IEEE, pp 4214–4217
Tan K, Wang D (2018) A convolutional recurrent neural network for real-time speech enhancement. In: Interspeech, pp 3229–3233
Tan K, Wang D (2019) Learning complex spectral mapping with gated convolutional recurrent networks for monaural speech enhancement. IEEE/ACM Trans Audio Speech Lang Process 28:380–390
Thiemann J, Ito N, Vincent E (2013) The diverse environments multi-channel acoustic noise database (demand): A database of multichannel environmental noise recordings. In: Proceedings of Meetings on Acoustics, AIP Publishing
Tigga NP, Garg S (2022) Efficacy of novel attention-based gated recurrent units transformer for depression detection using electroencephalogram signals. Health Inf Sci Syst 11(1):1
Ullah R, Wuttisittikulkij L, Chaudhary S et al (2022) End-to-end deep convolutional recurrent models for noise robust waveform speech enhancement. Sensors 22(20):7782
Valentini-Botinhao C, Wang X, Takaki S, et al (2016) Investigating rnn-based speech enhancement methods for noise-robust text-to-speech. In: SSW, pp 146–152
Vaswani A, Shazeer N, Parmar N, et al (2017) Attention is all you need. Adv Neural Inf Process Syst 30
Wang K, Cai J, Yao J et al (2021) Co-teaching based pseudo label refinery for cross-domain object detection. IET Image Process 15(13):3189–3199
Wang K, He B, Zhu WP (2021) Tstnn: Two-stage transformer based neural network for speech enhancement in the time domain. ICASSP 2021–2021 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 7098–7102
Wang SH, Fernandes SL, Zhu Z et al (2021) Avnc: attention-based vgg-style network for covid-19 diagnosis by cbam. IEEE Sens J 22(18):17431–17438
Wang Z, Zhang T, Shao Y et al (2021) Lstm-convolutional-blstm encoder-decoder network for minimum mean-square error approach to speech enhancement. Appl Acoust 172:107647
Xian Y, Sun Y, Wang W et al (2021) Convolutional fusion network for monaural speech enhancement. Neural Netw 143:97–107
Xian Y, Sun Y, Wang W, et al (2021) Multi-scale residual convolutional encoder decoder with bidirectional long short-term memory for single channel speech enhancement. In: 2020 28th European Signal Processing Conference (EUSIPCO), IEEE, pp 431–435
Xiang X, Zhang X, Chen H (2021) A nested u-net with self-attention and dense connectivity for monaural speech enhancement. IEEE Signal Process Lett 29:105–109
Xu S, Fosler-Lussier E (2019) Spatial and channel attention based convolutional neural networks for modeling noisy speech. ICASSP 2019–2019 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6625–6629
Yadav S, Rai A (2020) Frequency and temporal convolutional attention for text-independent speaker recognition. In: ICASSP 2020-2020 IEEE international conference on acoustics, speech and signal processing (ICASSP), IEEE, pp 6794–6798
Yamaguchi T, Ota J, Otake M (2012) A system that assists group conversation of older adults by evaluating speech duration and facial expression of each participant during conversation. In: 2012 IEEE International conference on robotics and automation, IEEE, pp 4481–4486
Yu F, Koltun V (2015) Multi-scale context aggregation by dilated convolutions. arXiv preprint arXiv:1511.07122
Yu G, Li A, Wang H et al (2022) Dbt-net: Dual-branch federative magnitude and phase estimation with attention-in-attention transformer for monaural speech enhancement. IEEE/ACM Trans Audio Speech Lang Process 30:2629–2644
Zhang Q, Nicolson A, Wang M, et al (2019) Monaural speech enhancement using a multi-branch temporal convolutional network. arXiv preprint arXiv:1912.12023
Zhao H, Zarar S, Tashev I et al (2018) Convolutional-recurrent neural networks for speech enhancement. 2018 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 2401–2405
Zhao S, Nguyen TH, Ma B (2021) Monaural speech enhancement with complex convolutional block attention module and joint time frequency losses. ICASSP 2021–2021 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), IEEE, pp 6648–6652
Funding
No Funding.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of Interest
The authors declared that they have no conflict of interest in this work.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Parisae, V., Bhavanam, S.N. Adaptive attention mechanism for single channel speech enhancement. Multimed Tools Appl (2024). https://doi.org/10.1007/s11042-024-19076-0
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11042-024-19076-0