mlsanityData to 19 Oct 2026

LMPAN: A Lightweight Multi-Path Alignment Network for Joint Full-Duplex Acoustic Echo Cancellation and Noise Suppression

arXiv 2607.02062 ↗First posted 2 Jul 2026eess.AScited by 0

On the map: Speech and audio language models → Speech enhancement and noise suppression

Abstract (arXiv)

We propose a lightweight multi-path alignment network (LMPAN) for on-device joint acoustic echo cancellation (AEC) and noise suppression (NS) in full-duplex spoken dialogue systems. To address hardware-induced distortions and dynamic acoustic conditions, we introduce three core innovations: (1) a multi-path alignment stage correcting temporal and energy mismatches across reference, linear AEC (LAEC) output, and microphone signals; (2) an attention-based mechanism that dynamically integrates enhanced LAEC and microphone features under varying acoustic scenarios; (3) a post-filtering module with a dynamic target generation strategy for downstream tasks (ASR, VAD). Furthermore, we adopt a two-stage training framework leveraging self-supervised learning representations to enhance perceptual quality. Experiments show that LMPAN, with only 480K parameters and 126 MACs, achieves performance comparable to the state-of-the-art lightweight model DeepVQE-S, while ensuring real-time inference capability.

Key facts. No citing papers in the map yet (data through 19 Oct 2026). Placed in Speech enhancement and noise suppression, one of 11 directions in Speech and audio language models. Builds on 23 of its 34 references that are in the map. First posted on arXiv 2 Jul 2026.

Builds on

23 of 34 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025A Small-footprint Acoustic Echo Cancellation Solution for Mobile Full-Duplex Speech Interactions11 Aug
  2. 2024A Closer Look at Wav2Vec2 Embeddings for On-Device Single-Channel Speech Enhancement3 Mar
  3. 2023ICASSP 2023 Acoustic Echo Cancellation Challenge22 Sep
  4. 2022SCA: Streaming Cross-attention Alignment for Echo Cancellation1 Nov
  5. 2022ICASSP 2022 Acoustic Echo Cancellation Challenge27 Feb

Other references · 18

  1. 2021WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processingcited by 2,135
  2. 2019SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognitioncited by 1,636
  3. 2020The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Testing Framework, and Challenge Resultscited by 230
  4. 2022Paraformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech Recognitioncited by 198
  5. 2022ICASSP 2022 Deep Noise Suppression Challengecited by 107
  6. 2024A Full-duplex Speech Dialogue Scheme Based On Large Language Modelscited by 51
  7. 2020Data augmentation and loss normalization for deep noise suppressioncited by 42
  8. 2023Vec-Tok Speech: speech vectorization and tokenization for neural speech generationcited by 20
  9. 2025LLM-Enhanced Dialogue Management for Full-Duplex Spoken Dialogue Systemscited by 16
  10. 2021AECMOS: A speech quality assessment metric for echo impairmentcited by 15
  11. 2023DeepVQE: Real Time Deep Voice Quality Enhancement for Joint Acoustic Echo Cancellation, Noise Suppression and Dereverberationcited by 12
  12. 2023Semantic VAD: Low-Latency Voice Activity Detection for Speech Interactioncited by 8
6 more
  1. 2022Multi-Task Deep Residual Echo Suppression with Echo-aware Losscited by 7
  2. 2020A study on more realistic room simulation for far-field keyword spottingcited by 5
  3. 2023Two-step Band-split Neural Network Approach for Full-band Residual Echo Suppressioncited by 3
  4. 2024Align-ULCNet: Towards Low-Complexity and Robust Acoustic Echo and Noise Reductioncited by 3
  5. 2025EchoFree: Towards Ultra Lightweight and Efficient Neural Acoustic Echo Cancellationcited by 2
  6. 2024FADI-AEC: Fast Score Based Diffusion Model Guided by Far-end Signal for Acoustic Echo Cancellationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2022Personalized Acoustic Echo Cancellation for Full-duplex Communications ↗
  2. 2022Task splitting for DNN-based acoustic echo and noise removal
  3. 2025SALMONN-omni: A Standalone Speech LLM without Codec Injection for Full-duplex Conversation
  4. 2023PDPCRN: Parallel Dual-Path CRN with Bi-directional Inter-Branch Interactions for Multi-Channel Speech Enhancement
  5. 2025FlexDuo: A Pluggable System for Enabling Full-Duplex Capabilities in Speech Dialogue Systems
  6. 2021Deep Residual Echo Suppression and Noise Reduction: A Multi-Input FCRN Approach in a Hybrid Speech Enhancement System
  7. 2021End-to-End Complex-Valued Multidilated Convolutional Neural Network for Joint Acoustic Echo Cancellation and Noise Suppression
  8. 2021A Conformer-based ASR Frontend for Joint Acoustic Echo Cancellation, Speech Enhancement and Speech Separation