Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A two-step approach for speech enhancement in low-SNR scenarios using cyclostationary beamforming and DNNs

Giovanni Bologni, Nicolás Arrieta Larraza, Richard Heusdens, Richard C. Hendriks

TL;DR

Results indicate that explicitly incorporating cyclostationarity as a signal prior is more effective than increasing model capacity alone for suppressing harmonic interference.

Abstract

Deep Neural Networks (DNNs) often struggle to suppress noise at low signal-to-noise ratios (SNRs). This paper addresses speech enhancement in scenarios dominated by harmonic noise and proposes a framework that integrates cyclostationarity-aware preprocessing with lightweight DNN-based denoising. A cyclic minimum power distortionless response (cMPDR) spectral beamformer is used as a preprocessing block. It exploits the spectral correlations of cyclostationary noise to suppress harmonic components prior to learning-based enhancement and does not require modifications to the DNN architecture. The proposed pipeline is evaluated in a single-channel setting using two DNN architectures: a simple and lightweight convolutional recurrent neural network (CRNN), and a state-of-the-art model, namely ultra-low complexity network (ULCNet). Experiments on synthetic data and real-world recordings dominated by rotating machinery noise demonstrate consistent improvements over end-to-end DNN baselines, particularly at low SNRs. Remarkably, a parameter-efficient CRNN with cMPDR preprocessing surpasses the performance of the larger ULCNet operating on raw or Wiener-filtered inputs. These results indicate that explicitly incorporating cyclostationarity as a signal prior is more effective than increasing model capacity alone for suppressing harmonic interference.

A two-step approach for speech enhancement in low-SNR scenarios using cyclostationary beamforming and DNNs

TL;DR

Results indicate that explicitly incorporating cyclostationarity as a signal prior is more effective than increasing model capacity alone for suppressing harmonic interference.

Abstract

Deep Neural Networks (DNNs) often struggle to suppress noise at low signal-to-noise ratios (SNRs). This paper addresses speech enhancement in scenarios dominated by harmonic noise and proposes a framework that integrates cyclostationarity-aware preprocessing with lightweight DNN-based denoising. A cyclic minimum power distortionless response (cMPDR) spectral beamformer is used as a preprocessing block. It exploits the spectral correlations of cyclostationary noise to suppress harmonic components prior to learning-based enhancement and does not require modifications to the DNN architecture. The proposed pipeline is evaluated in a single-channel setting using two DNN architectures: a simple and lightweight convolutional recurrent neural network (CRNN), and a state-of-the-art model, namely ultra-low complexity network (ULCNet). Experiments on synthetic data and real-world recordings dominated by rotating machinery noise demonstrate consistent improvements over end-to-end DNN baselines, particularly at low SNRs. Remarkably, a parameter-efficient CRNN with cMPDR preprocessing surpasses the performance of the larger ULCNet operating on raw or Wiener-filtered inputs. These results indicate that explicitly incorporating cyclostationarity as a signal prior is more effective than increasing model capacity alone for suppressing harmonic interference.
Paper Structure (16 sections, 6 equations, 2 figures, 1 table)

This paper contains 16 sections, 6 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Proposed method: cMPDR-DNN
  3. cMPDR preprocessor
  4. Deep learning architectures
  5. CRNN
  6. ULCNet ()
  7. Experimental setup
  8. Datasets
  9. Synthetic dataset
  10. Real-world dataset
  11. Baselines, performance metrics
  12. Algorithm implementation and training details
  13. Results
  14. Evaluation on synthetic data
  15. Evaluation on real-world data
  16. ...and 1 more sections

Figures (2)

  • Figure 1: Mel spectrograms of a recording from the IDMT-ISA Electric Engine dataset, showing noisy input, clean target, and outputs of the CRNN and proposed CRNN+cMPDR pipeline. The highlights show that CRNN+cMPDR can recover heavily masked speech components, illustrating the benefits of cyclostationarity-aware preprocessing.
  • Figure 2: Mean values of the objective metrics as a function of the SNR using the synthetic harmonic noise dataset.