Listen first: Output-based multi-microphone speech enhancement

Traditionally, hearing-aid speech enhancement (SE) algorithms rely on input-based feature estimation, often derived by a voice activity detector (VAD), to configure beamformers. Yet features extracted from noisy microphone signals can become unreliable in challenging acoustic scenes where users most need help. We introduce a novel paradigm in which the settings of a sound processing system are determined by evaluating characteristics of its output. To demonstrate this idea, we employ an output-based system that selects among a set of minimum power distortionless response (MPDR) beamformers. Although MPDR beamformers are typically avoided due to their sensitivity to steering errors, we show that they become effective within an output-based framework. We compare the proposed system to a conventional input-based minimum variance distortionless response (MVDR) baseline. Experimental results show that the proposed system consistently outperforms the MVDR baseline, particularly at low SNRs, in terms of SNR, ESTOI and PESQ.

An Objective Intelligibility Metric Evaluation on Spanish Speech

Objective intelligibility metrics (OIMs) enable fast and low-cost evaluation of speech intelligibility and are widely used in speech technology assessment. In this study, we evaluate five reference-based OIMs (STOI, ESTOI, STGI, HASPI, and SIIB) and two deep learning-based no-reference metrics (MOSA-Net+ and W2V-SIP) on SpInt, a new Spanish speech intelligibility dataset. Our results show that reference-based OIMs consistently outperform modern data-driven no-reference approaches, which degrade notably under training-test acoustic mismatches such as language mismatch. This effect is particularly relevant in our scenario, as none of the evaluated metrics were exposed to Spanish speech data during development. Consequently, to foster research on more robust and generalizable no-reference OIMs, SpInt is released publicly.

Ranking the Impact of Contextual Specialization in Neural Speech Enhancement

We systematically investigate neural speech enhancement systems, ranging from very small ($\sim$10\,k parameters) to medium-large ($\sim$2-5\,M parameters), which specialize to acoustic conditions using contextual information such as speaker identity, noise type, speaker gender, spoken language, and SNR. By fine-tuning generalist models on specific data subsets, we find that specializing to a speaker's identity consistently yields the largest gains in estimated speech intelligibility and quality. In contrast, specializing to SNR, noise type, or gender offers only marginal benefits. Crucially, we show that a small model specialized to both a specific speaker and a specific noise type can match or exceed the performance of a generalist model ten times its size. Further, cross-lingual tests reveal that models specialized to a target language outperform multilingual generalists, suggesting that language is a salient feature for specialization. These findings highlight the potential of small, adaptive models for resource-constrained applications like hearing aids, which specialize on-the-fly to contextual information.

Exploring Resolution-Wise Shared Attention in Hybrid Mamba-U-Nets for Improved Cross-Corpus Speech Enhancement

Recent advances in speech enhancement have shown that models combining Mamba and attention mechanisms yield superior cross-corpus generalization performance. At the same time, integrating Mamba in a U-Net structure has yielded state-of-the-art enhancement performance, while reducing both model size and computational complexity. Inspired by these insights, we propose RWSA-MambaUNet, a novel and efficient hybrid model combining Mamba and multi-head attention in a U-Net structure for improved cross-corpus performance. Resolution-wise shared attention (RWSA) refers to layerwise attention-sharing across corresponding time- and frequency resolutions. Our best-performing RWSA-MambaUNet model achieves state-of-the-art generalization performance on two out-of-domain test sets. Notably, our smallest model surpasses all baselines on the out-of-domain DNS 2020 test set in terms of PESQ, SSNR, and ESTOI, and on the out-of-domain EARS-WHAM_v2 test set in terms of SSNR, ESTOI, and SI-SDR, while using less than half the model parameters and a fraction of the FLOPs.

Learning Robust Spatial Representations from Binaural Audio through Feature Distillation

Recently, deep representation learning has shown strong performance in multiple audio tasks. However, its use for learning spatial representations from multichannel audio is underexplored. We investigate the use of a pretraining stage based on feature distillation to learn a robust spatial representation of binaural speech without the need for data labels. In this framework, spatial features are computed from clean binaural speech samples to form prediction labels. These clean features are then predicted from corresponding augmented speech using a neural network. After pretraining, we throw away the spatial feature predictor and use the learned encoder weights to initialize a DoA estimation model which we fine-tune for DoA estimation. Our experiments demonstrate that the pretrained models show improved performance in noisy and reverberant environments after fine-tuning for direction-of-arrival estimation, when compared to fully supervised models and classic signal processing methods.

Head-steered channel selection method for hearing aid applications using remote microphones

We propose a channel selection method for hearing aid applications using remote microphones, in the presence of multiple competing talkers. The proposed channel selection method uses the hearing aid user's head-steering direction to identify the remote channel originating from the frontal direction of the hearing aid user, which captures the target talker signal. We pose the channel selection task as a multiple hypothesis testing problem, and derive a maximum likelihood solution. Under realistic, simplifying assumptions, the solution selects the remote channel which has the highest weighted squared absolute correlation coefficient with the output of the head-steered hearing aid beamformer. We analyze the performance of the proposed channel selection method using close-talking remote microphones and table microphone arrays. Through simulations using realistic acoustic scenes, we show that the proposed channel selection method consistently outperforms existing methods in accurately finding the remote channel that captures the target talker signal, in the presence of multiple competing talkers, without the use of any additional sensors.

Binaural Localization Model for Speech in Noise

Binaural acoustic source localization is important to human listeners for spatial awareness, communication and safety. In this paper, an end-to-end binaural localization model for speech in noise is presented. A lightweight convolutional recurrent network that localizes sound in the frontal azimuthal plane for noisy reverberant binaural signals is introduced. The model incorporates additive internal ear noise to represent the frequency-dependent hearing threshold of a typical listener. The localization performance of the model is compared with the steered response power algorithm, and the use of the model as a measure of interaural cue preservation for binaural speech enhancement methods is studied. A listening test was performed to compare the performance of the model with human localization of speech in noisy conditions.

MambAttention: Mamba with Multi-Head Attention for Generalizable Single-Channel Speech Enhancement

With the advent of new sequence models like Mamba and xLSTM, several studies have shown that these models match or outperform state-of-the-art models in single-channel speech enhancement, automatic speech recognition, and self-supervised audio representation learning. However, prior research has demonstrated that sequence models like LSTM and Mamba tend to overfit to the training set. To address this issue, previous works have shown that adding self-attention to LSTMs substantially improves generalization performance for single-channel speech enhancement. Nevertheless, neither the concept of hybrid Mamba and time-frequency attention models nor their generalization performance have been explored for speech enhancement. In this paper, we propose a novel hybrid architecture, MambAttention, which combines Mamba and shared time- and frequency-multi-head attention modules for generalizable single-channel speech enhancement. To train our model, we introduce VoiceBank+Demand Extended (VB-DemandEx), a dataset inspired by VoiceBank+Demand but with more challenging noise types and lower signal-to-noise ratios. Trained on VB-DemandEx, our proposed MambAttention model significantly outperforms existing state-of-the-art LSTM-, xLSTM-, Mamba-, and Conformer-based systems of similar complexity across all reported metrics on two out-of-domain datasets: DNS 2020 and EARS-WHAM_v2, while matching their performance on the in-domain dataset VB-DemandEx. Ablation studies highlight the role of weight sharing between the time- and frequency-multi-head attention modules for generalization performance. Finally, we explore integrating the shared time- and frequency-multi-head attention modules with LSTM and xLSTM, which yields a notable performance improvement on the out-of-domain datasets. However, our MambAttention model remains superior on both out-of-domain datasets across all reported evaluation metrics.

xLSTM-SENet: xLSTM for Single-Channel Speech Enhancement

While attention-based architectures, such as Conformers, excel in speech enhancement, they face challenges such as scalability with respect to input sequence length. In contrast, the recently proposed Extended Long Short-Term Memory (xLSTM) architecture offers linear scalability. However, xLSTM-based models remain unexplored for speech enhancement. This paper introduces xLSTM-SENet, the first xLSTM-based single-channel speech enhancement system. A comparative analysis reveals that xLSTM-and notably, even LSTM-can match or outperform state-of-the-art Mamba- and Conformer-based systems across various model sizes in speech enhancement on the VoiceBank+Demand dataset. Through ablation studies, we identify key architectural design choices such as exponential gating and bidirectionality contributing to its effectiveness. Our best xLSTM-based model, xLSTM-SENet2, outperforms state-of-the-art Mamba- and Conformer-based systems on the Voicebank+DEMAND dataset.

Noise-Robust Target-Speaker Voice Activity Detection Through Self-Supervised Pretraining

Target-Speaker Voice Activity Detection (TS-VAD) is the task of detecting the presence of speech from a known target-speaker in an audio frame. Recently, deep neural network-based models have shown good performance in this task. However, training these models requires extensive labelled data, which is costly and time-consuming to obtain, particularly if generalization to unseen environments is crucial. To mitigate this, we propose a causal, Self-Supervised Learning (SSL) pretraining framework, called Denoising Autoregressive Predictive Coding (DN-APC), to enhance TS-VAD performance in noisy conditions. We also explore various speaker conditioning methods and evaluate their performance under different noisy conditions. Our experiments show that DN-APC improves performance in noisy conditions, with a general improvement of approx. 2% in both seen and unseen noise. Additionally, we find that FiLM conditioning provides the best overall performance. Representation analysis via tSNE plots reveals robust initial representations of speech and non-speech from pretraining. This underscores the effectiveness of SSL pretraining in improving the robustness and performance of TS-VAD models in noisy environments.