mlsanityData to 19 Oct 2026

μNet: Ultra-Low-Memory and Low-Complexity Speech Enhancement for Embedded Digital Signal Processors

arXiv 2608.21155 ↗First posted 21 Aug 2026eess.AScited by 0

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

Abstract (arXiv)

Speech enhancement on embedded digital signal processors (DSPs) imposes strict constraints on memory footprint, computational complexity, latency, and support for integer operations. Although recent DNN-based approaches have addressed these challenges individually, no unified framework in the literature simultaneously addresses all these requirements for practical deployment. In this work, we propose μNet, an ultra-low-memory, low-complexity, and low-latency end-to-end DNN model. The proposed method requires only 90~KB of static memory and 28~MMACs, while supporting an algorithmic latency as low as 4~ms with performance comparable to state-of-the-art methods of similar complexity. Our experiments demonstrate that μNet is compatible with neural accelerators and supports full integer-arithmetic operations on consumer DSP platforms such as Cadence Tensilica HiFi 4/5.

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 27 of its 40 references that are in the map. First posted on arXiv 21 Aug 2026.

Builds on

27 of 40 references.

Closest ancestors · 11

Same direction, newest first

  1. 2026A two-step approach for speech enhancement in low-SNR scenarios using cyclostationary beamforming and DNNs13 Feb
  2. 2026Fast-ULCNet: A fast and ultra low complexity network for single-channel speech enhancement21 Jan
  3. 2024Modulating State Space Model with SlowFast Framework for Compute-Efficient Ultra Low-Latency Speech Enhancement4 Nov
  4. 2024Ultra-Low Latency Speech Enhancement - A Comprehensive Study16 Sep
  5. 2024Rethinking Processing Distortions: Disentangling the Impact of Speech Enhancement Errors on Speech Recognition Performance23 Apr
  6. 2023Ultra Low Complexity Deep Learning Based Noise Suppression13 Dec
  7. 2022DeepFilterNet2: Towards Real-Time Speech Enhancement on Embedded Devices for Full-Band Audio11 May
  8. 2022STFT-Domain Neural Speech Enhancement with Very Low Algorithmic Latency21 Apr
  9. 2021DeepFilterNet: A Low Complexity Speech Enhancement Framework for Full-Band Audio based on Deep Filtering11 Oct
  10. 2021Real-time Denoising and Dereverberation with Tiny Recurrent U-Net5 Feb
  11. 2021Towards efficient models for real-time deep noise suppression22 Jan

Other references · 16

  1. 2017Attention Is All You Needcited by 61,180
  2. 2018ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Designcited by 1,143
  3. 2018SDR - half-baked or well done?cited by 626
  4. 2018Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separationcited by 596
  5. 2021DNSMOS P.835: A Non-Intrusive Perceptual Objective Speech Quality Metric to Evaluate Noise Suppressorscited by 292
  6. 2020DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancementcited by 284
  7. 2017Parallel WaveNet: Fast High-Fidelity Speech Synthesiscited by 240
  8. 2020The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Testing Framework, and Challenge Resultscited by 230
  9. 2020TensorFlow Lite Micro: Embedded Machine Learning on TinyML Systemscited by 171
  10. 2017A Hybrid DSP/Deep Learning Approach to Real-Time Full-Band Speech Enhancementcited by 67
  11. 2020A Perceptually-Motivated Approach for Low-Complexity, Real-Time Enhancement of Fullband Speechcited by 53
  12. 2020Channel-wise Subband Input for Better Voice and Accompaniment Separation on High Resolution Musiccited by 15
4 more
  1. 2021Deep neural network Based Low-latency Speech Separation with Asymmetric analysis-Synthesis Window Paircited by 10
  2. 2022Low-latency Monaural Speech Enhancement with Deep Filter-bank Equalizercited by 5
  3. 2024A Hybrid Approach for Low-Complexity Joint Acoustic Echo and Noise Reductioncited by 4
  4. 2024Align-ULCNet: Towards Low-Complexity and Robust Acoustic Echo and Noise Reductioncited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2023Single Channel Speech Enhancement Using U-Net Spiking Neural Networks
  2. 2026Beyond Model Size: Redesigning LiSenNet for embedded speech enhancement
  3. 2026Feasibility of Time-Domain DNN-Based Speech Enhancement on Embedded FPGA for Hearing Aids
  4. 2022Accelerating RNN-based Speech Enhancement on a Multi-Core MCU with Mixed FP16-INT8 Post-Training Quantization
  5. 2021Weight, Block or Unit? Exploring Sparsity Tradeoffs for Speech Enhancement on Tiny Neural Accelerators
  6. 2025A Low-Power Streaming Speech Enhancement Accelerator For Edge Devices
  7. 2026Does per-frame early exit pay? A compute-matched study of dynamic depth for on-device speech enhancement
  8. 2025UL-UNAS: Ultra-Lightweight U-Nets for Real-Time Speech Enhancement via Network Architecture Search