Research map / Architectures and efficiency / Spiking networks
DBHN-Net: Dual-Branch Hybrid Neural Network for Low-Complexity Monaural Speech Enhancement
Cunhang Fan, Enrui Liu, Jin Zhou, Jian Kang, Jie Li, Andong Li, Jin Zhou, Zhao Lv, Xuelong Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the ANN branch to the hybrid network (ablation) | DBHN-Net · WSJ0-SI84+DNS (0 dB) | PESQ: 2.75 -> 3.17 | Improved |
| add the SNN branch to the hybrid network (ablation) | DBHN-Net · WSJ0-SI84+DNS (0 dB) | PESQ: 2.81 -> 3.17 | Improved |
| use Mamba instead of LSTM for sequence modeling (ablation) | DBHN-Net ANN branch · WSJ0-SI84+DNS (0 dB) | PESQ / MACs: 3.01 -> 3.17; 8.68G/s -> 1.32G/s | Improved |
| use Mamba instead of Transformer for sequence modeling (ablation) | DBHN-Net ANN branch · WSJ0-SI84+DNS (0 dB) | PESQ / MACs: 3.13 -> 3.17; 22.05G/s -> 1.32G/s | Improved |
| use SFEB instead of a simple LIF unit (ablation) | DBHN-Net SNN branch · WSJ0-SI84+DNS (0 dB) | SI-SDR: 11.33 -> 12.16 | Improved |
| add the Information Transformation Block (ablation) | DBHN-Net SNN branch · WSJ0-SI84+DNS (0 dB) | PESQ: 3.11 -> 3.17 | Improved |
| add the Interaction module (ablation) | DBHN-Net · WSJ0-SI84+DNS (0 dB) | PESQ: 3.08 -> 3.17 | Improved |
| add the TF-CAF cross-attention fusion module (ablation) | DBHN-Net · WSJ0-SI84+DNS (0 dB) | ESTOI: 79.19 -> 83.46 | Improved |
| add both Interaction and TF-CAF fusion modules (ablation) | DBHN-Net · WSJ0-SI84+DNS (0 dB) | PESQ: 2.91 -> 3.17 | Improved |
| add an SNN branch to a Mamba dual-branch ANN model (vs BSDBNet) | DBHN-Net vs BSDBNet · VoiceBank+Demand | WB-PESQ / MACs: 3.17 -> 3.08; 1.68 G/s -> 1.32G/s | Mixed |
| hybrid ANN-SNN design vs best prior SNN model | DBHN-Net vs Spiking-UNet · WSJ0-SI84+DNS (avg SNR) | PESQ: 2.44 -> 3.14 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- BSDB-Net — BSDB-Net: Band-Split Dual-Branch Network with Selective State Spaces Mechanism for Monaural Speech Enhancement (AAAI 2025) · better on 5 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- BSDB-Net: Band-Split Dual-Branch Network with Selective State Spaces Mechanism for Monaural Speech Enhancement (AAAI 2025 · similarity 0.88)
- EnOF-SNN: Training Accurate Spiking Neural Networks via Enhancing the Output Feature (NeurIPS 2024 · similarity 0.84)
- Balanced Resonate-and-Fire Neurons (ICML 2024 · similarity 0.84)
- FINALLY: fast and universal speech enhancement with studio-like quality (NeurIPS 2024 · similarity 0.84)
- Efficient ANN-Guided Distillation: Aligning Rate-based Features of Spiking Neural Networks through Hybrid Block-wise Replacement (CVPR 2025 · similarity 0.83)
- LM-HT SNN: Enhancing the Performance of SNN to ANN Counterpart through Learnable Multi-hierarchical Threshold Model (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
