Fast-ULCNet: A fast and ultra low complexity network for single-channel speech enhancement

Single-channel speech enhancement algorithms are often used in resource-constrained embedded devices, where low latency and low complexity designs gain more importance. In recent years, researchers have proposed a wide variety of novel solutions to this problem. In particular, a recent deep learning model named ULCNet is among the state-of-the-art approaches in this domain. This paper proposes an adaptation of ULCNet, by replacing its GRU layers with FastGRNNs, to reduce both computational latency and complexity. Furthermore, this paper shows empirical evidence on the performance decay of FastGRNNs in long audio signals during inference due to internal state drifting, and proposes a novel approach based on a trainable complementary filter to mitigate it. The resulting model, Fast-ULCNet, performs on par with the state-of-the-art original ULCNet architecture on a speech enhancement task, while reducing its model size by more than half and decreasing its latency by 34% on average.

• Publication year

  2026

• Venue

  arXiv.org

• Publication date

  2026-01-21

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.48550/arXiv.2601.14925arXiv 2601.14925
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Sixty Years of Frequency-Domain Monaural Speech Enhancement: From Traditional to Deep Learning Methods

  2023cited by this paper
• Low Complexity Speech Enhancement Network Based on Frame-Level Swin Transformer

  2023cited by this paper
• Ultra Low Complexity Deep Learning Based Noise Suppression

  2023influential reference
• Deepfilternet: A Low Complexity Speech Enhancement Framework for Full-Band Audio Based On Deep Filtering

  2021cited by this paper
• Real-Time Denoising and Dereverberation wtih Tiny Recurrent U-Net

  2021cited by this paper
• Dnsmos P.835: A Non-Intrusive Perceptual Objective Speech Quality Metric to Evaluate Noise Suppressors

  2021cited by this paper
• A consolidated view of loss functions for supervised deep learning-based speech enhancement

  2020cited by this paper
• The INTERSPEECH 2020 Deep Noise Suppression Challenge: Datasets, Subjective Testing Framework, and Challenge Results

  2020cited by this paper
• FastGRNN: A Fast, Accurate, Stable and Tiny Kilobyte Sized Gated Recurrent Neural Network

  2018influential reference
• SDR – Half-baked or Well Done?

  2018cited by this paper
• Complex Ratio Masking for Monaural Speech Separation

  2016cited by this paper
• On the Properties of Neural Machine Translation: Encoder–Decoder Approaches

  2014cited by this paper
• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

  2014cited by this paper

• No citing papers are available for this paper.