Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SS-BRPE: Self-Supervised Blind Room Parameter Estimation Using Attention Mechanisms

Chunxi Wang, Maoshen Jia, Meiran Li, Changchun Bao, Wenyu Jin

TL;DR

This work tackles blind estimation of room geometry and reverberation (volume and RT$_{60}$) from single-channel speech without relying on costly labeled data. It introduces SS-BRPE, a purely attention-based model built on an Audio Spectrogram Transformer and pre-trained with self-supervised discriminative and generative patch tasks on unlabeled audio, followed by fine-tuning with dynamic online feature augmentation. Empirical results show SS-BRPE outperforms CNN/CRNN and ImageNet-pretrained attention baselines in both volume and RT$_{60}$ estimation, and remains robust when labeled data are scarce. The method reduces data labeling requirements and offers practical benefits for real-world acoustic parameter estimation and downstream audio processing tasks, with potential for broader applicability in blind room parameter inference.

Abstract

In recent years, dynamic parameterization of acoustic environments has garnered attention in audio processing. This focus includes room volume and reverberation time (RT60), which define local acoustics independent of sound source and receiver orientation. Previous studies show that purely attention-based models can achieve advanced results in room parameter estimation. However, their success relies on supervised pretrainings that require a large amount of labeled true values for room parameters and complex training pipelines. In light of this, we propose a novel Self-Supervised Blind Room Parameter Estimation (SS-BRPE) system. This system combines a purely attention-based model with self-supervised learning to estimate room acoustic parameters, from single-channel noisy speech signals. By utilizing unlabeled audio data for pretraining, the proposed system significantly reduces dependencies on costly labeled datasets. Our model also incorporates dynamic feature augmentation during fine-tuning to enhance adaptability and generalizability. Experimental results demonstrate that the SS-BRPE system not only achieves more superior performance in estimating room parameters than state-of-the-art (SOTA) methods but also effectively maintains high accuracy under conditions with limited labeled data. Code available at https://github.com/bjut-chunxiwang/SS-BRPE.

SS-BRPE: Self-Supervised Blind Room Parameter Estimation Using Attention Mechanisms

TL;DR

This work tackles blind estimation of room geometry and reverberation (volume and RT) from single-channel speech without relying on costly labeled data. It introduces SS-BRPE, a purely attention-based model built on an Audio Spectrogram Transformer and pre-trained with self-supervised discriminative and generative patch tasks on unlabeled audio, followed by fine-tuning with dynamic online feature augmentation. Empirical results show SS-BRPE outperforms CNN/CRNN and ImageNet-pretrained attention baselines in both volume and RT estimation, and remains robust when labeled data are scarce. The method reduces data labeling requirements and offers practical benefits for real-world acoustic parameter estimation and downstream audio processing tasks, with potential for broader applicability in blind room parameter inference.

Abstract

In recent years, dynamic parameterization of acoustic environments has garnered attention in audio processing. This focus includes room volume and reverberation time (RT60), which define local acoustics independent of sound source and receiver orientation. Previous studies show that purely attention-based models can achieve advanced results in room parameter estimation. However, their success relies on supervised pretrainings that require a large amount of labeled true values for room parameters and complex training pipelines. In light of this, we propose a novel Self-Supervised Blind Room Parameter Estimation (SS-BRPE) system. This system combines a purely attention-based model with self-supervised learning to estimate room acoustic parameters, from single-channel noisy speech signals. By utilizing unlabeled audio data for pretraining, the proposed system significantly reduces dependencies on costly labeled datasets. Our model also incorporates dynamic feature augmentation during fine-tuning to enhance adaptability and generalizability. Experimental results demonstrate that the SS-BRPE system not only achieves more superior performance in estimating room parameters than state-of-the-art (SOTA) methods but also effectively maintains high accuracy under conditions with limited labeled data. Code available at https://github.com/bjut-chunxiwang/SS-BRPE.
Paper Structure (13 sections, 3 figures, 2 tables)

This paper contains 13 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Model architecture
  3. Self-Supervised Blind Room Parameter Estimation Model
  4. Audio Spectrogram Transformer
  5. Self-supervised Learning Framework
  6. Pretraining
  7. Fine-tuning with Feature Augmention
  8. Experiments
  9. Experimental Design
  10. Experimental Results
  11. Estimation of Room Volume & RT$_{60}$
  12. Estimation of Room Parameters with Limited Data
  13. Conclusion and Future work

Figures (3)

  • Figure 1: The Self-Supervised Blind Room Parameter Estimation (SS-BRPE) architecture.
  • Figure 2: Patch masking operation in online feature augmentation. Black patches represent random rectangular masks applied to 2-D audio feature blocks.
  • Figure 3: Histograms of room volume and RT$_{60}$ distributions across various datasets. The horizontal axis represents scales of room parameters and the vertical axis represents numbers of rooms.