1 School of Information and Communication, Guilin University of Electronic Technology, Guilin 541004, China
2 Key Laboratory of Cognitive Radio and Information Processing, Ministry of Education (Guilin University of Electronic Technology), Guilin 541004, China
3 GUET-Nanning E-Tech Research Institute Co., Ltd., Nanning 530000, China
In recent years, deep learning methods have made significant progress in the field of speech enhancement, but the existing models generally have the problems of redundant parameters and high inference overhead, which are difficult to meet the requirements of edge computing devices for low resource occupation and real-time performance. In order to solve this problem, an efficient and lightweight speech enhancement framework was proposed. The front-end of the model used the band merging (BM) to compress high-frequency redundant information. Group fusion convolution (GF-Conv) was introduced, and the dilated deep convolution and time-frequency hybrid dilated residual attention module (TF-HDRA) were combined to improve the multi-scale modeling ability. At the same time, the cross-channel feature interaction was enhanced by the channel aggregation (CA) module. The grouped dual-path long short-term memory (GDPLSTM) module was used in the enhancement layer, which combined channel grouping and dual-path mechanism, which not only maintained excellent time-frequency modeling capabilities, but also realized the lightweight of the model structure, and effectively reduced the number of parameters and inference overhead. Experiments were carried out on the VoiceBank+DEMAND dataset. The results show that the voice perception quality of the proposed model is 2.99, the parameter quantity is 51 850, and the computational amount is 145.23 MMAC, which achieves a good balance between performance and computing efficiency.
ZHENG Zhan-heng, XU Jia-yu, WANG Jian, LI Qi.
Lightweight Speech Enhancement Method Based on Grouped Dual-path LSTM[J].
Science Technology and Engineering, 2026, 26(20): 8712-8719 DOI:10.12404/j.issn.1671-1815.2505745
PanGuimei. Continuous speech recognition and pronunciation correction algorithm based on DTW and Transformer[J]. Modern Electronic Technique, 2025, 48(24): 61-66.
GongZhizi, MuYi, JiangWen, et al. Feasibility study of hearing aid howling based on AI speech enhancement algorithm[J]. Science Technology and Engineering, 2025, 42(9): 397-402.
LiJingying. Interactive speech noise reduction for translation robots based on fuzzy theory and convolutional neural network[J]. Automation & Instrumentation, 2025(1): 286-289, 294.
WangShiqi, ZengQingning, LongChao, et al. Multi-task learning for speech enhancement and tetection[J]. Computer Engineering and Applications, 2021, 57(20): 197-202.
[13]
MahmmodB M, AbdulhussainS H, AliT M, et al. Speechenhancement: a review of various approaches, trends, and challenges[C]// 17th International Conference on Development in eSystem Engineering (DeSE). New York: IEEE, 2024: 31-36.
[14]
O'ShaughnessyD. Speech enhancement—a review of modern methods[J]. IEEE Transactions on Human-Machine Systems, 2024, 54(1): 110-120.
[15]
PascualS, BonafonteA, SerraJ. SEGAN: speech enhancement generative adversarial network[C]// Proceedings of the 2017 Interspeech. Piscataway, NJ: IEEE, 2017: 3642-3646.
[16]
StollerD, EwertS, DixonS. Wave-U-Net: a multi-scale neural network for end-to-end audio source separation[EB/OL]. (2018-06-08)[2025-06-14]. https://arxiv.org/abs/1806.03185.
[17]
HuY X, LiuY, LvS B, et al. DCCRN: deep complex convolution recurrent network for phase-aware speech enhancement[C]// Proceedings of the 2020 Interspeech. Piscataway, NJ: IEEE Press, 2020: 2472-2476.
[18]
CaoR Z, AbdulatifS, YangB. CMGAN: conformer-based metric GAN for speech enhancement[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2024, 32: 2477-2493.
[19]
LiA D, ZhengC S, ZhangL, et al. Glance and gaze: a collaborative learning framework for single-channel speech enhancement[J]. Applied Acoustics, 2022, 187: 108499.
[20]
ZhaoS K, MaB. Mossformer: pushing the performance limit of monaural speech separation using gated single-head transformer with convolution-augmented joint self-attentions[C]// ICASSP 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Piscataway, NJ: IEEE, 2023: 1-5.
[21]
ValinJ M. A hybrid DSP/deep learning approach to real-time full-band speech enhancement[C]// IEEE 20th International Workshop on Multimedia Signal Processing (MMSP). New York: IEEE, 2018: 1-5.
[22]
SchroterH, RosenkranzT, et al. DeepFilterNet: a low complexity speech enhancement framework for full-band audio based on deep filtering[C]// ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). New York: IEEE, 2022: 7407-7411.
[23]
MaN N, ZhangX Y, ZhengH T, et al. Shufflenet v2: practical guidelines for efficient cnn architecture design[C]// Proceedings of the European Conference on Computer Vision (ECCV). Berlin: Springer, 2018: 116-131.
[24]
TanK, WangD L. Learning complex spectral mapping with gated convolutional recurrent networks for monaural speech enhancement[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2019, 28: 380-390.
[25]
LüS B, HuY X, ZhangS M, et al. DCCRN+: channel-wise subband dccrn with snr estimation for speech enhancement[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2021, 30: 1234-1245.
[26]
LüS B, FuY H, XingM T, et al. S-DCCRN: super wide band DCCRN with learnable complex feature for speech enhancement[C]// 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Piscataway, NJ: IEEE, 2022: 7767-7771.
[27]
ChenJ, WangZ L, TuoD Y, et al. FullSubNet+: channel attention FullSubNet with complex spectrograms for speech enhancement[C]// 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Piscataway, NJ: IEEE, 2022: 7857-7861.
[28]
ValinJM, IsikU, PhansalkarN, et al. A perceptually-motivated approach for low-complexity, real-time enhancement of fullband speech[C]// Interspeech 2020. Shanghai: ISCA, 2020: 2482-2486.