Audio--Image Alignment as a Continued-Pretraining Stage Improves Low-Resource ASR

Thousands of languages are spoken worldwide, yet many remain under-resourced for Automatic Speech Recognition (ASR) due to the limited availability of high-quality transcribed speech data. Collecting accurate transcriptions is often costly and labor-intensive, particularly for low-resource languages. In this work, we investigate the use of aligned audio-image pairs to adapt pretrained audio encoders without requiring transcription data before supervised fine-tuning. Our proposed representation alignment stage is introduced between large-scale pretraining and supervised ASR fine-tuning. Specifically, image representations extracted from pretrained vision encoders are aligned with audio representations to further adapt a pretrained audio encoder. For this alignment process, we utilize the Vaani dataset, in which images serve as prompts for speech collection, naturally providing paired audio-image data. We evaluate the proposed approach using multiple vision encoders and a pretrained FastConformer audio encoder. Experimental results demonstrate that models fine-tuned after representation alignment consistently achieve improved ASR performance compared to direct fine-tuning. These findings highlight the potential of audio-image representation alignment as an effective transcription-free adaptation strategy for enhancing ASR systems in low-resource language settings.

Joint Learning of Covariance Estimation and White Noise Gain for Robust MVDR Beamforming

The minimum variance distortionless response (MVDR) beamformer is widely used for multichannel speech enhancement due to strong noise suppression while preserving target signals. In practice, its performance is sensitive to microphone self-noise and array mismatches. Existing approaches typically rely on fixed, manually tuned WNG thresholds or diagonal loading, leading to suboptimal performance under unknown or time-varying acoustic conditions. This paper proposes a data-driven MVDR framework that adaptively estimates the WNG constraint using a deep neural network. The network jointly predicts a time-frequency noise mask for covariance estimation and a frequency-dependent WNG threshold, enabling dynamic robustness-directivity control. A differentiable robust MVDR layer is integrated into the framework, allowing end-to-end optimization. Experiments demonstrate consistent improvements in speech quality and intelligibility over conventional fixed-WNG MVDR methods.

Breaking Shortcut Learning for Cross-Trial EEG-Guided Target Speech Extraction via Two-Stage Training

Recent end-to-end models for EEG-guided target speech extraction report impressive results, underscoring potential for neuro-steered hearing technologies. However, our analysis reveals that high within-trial performance can be driven by trial-specific EEG structure that acts as shortcuts for target selection, leading to poor generalization on unseen trials. To overcome this gap, we propose TRUST-TSE, a two-stage framework to mitigate shortcut learning. By introducing contrastive pretraining with attended-speaker negative sampling, we encourage the EEG encoder to capture fine-grained EEG--speech alignment while suppressing trial-identity cues. We also employ a confidence-weighted extraction objective based on EEG--source similarity to guide extraction using the learned representations. Experiments on KUL and DTU datasets show that TRUST-TSE outperforms end-to-end baselines under strict cross-trial protocols, addressing a key reliability bottleneck of existing approaches.

ParaPairAudioBench: Paralinguistic Pairwise Audio Benchmark for LALM-as-a-Judge

Large Audio-Language Models (LALMs) have been widely used as judge models for the automatic evaluation of generated speech. However, prior approaches predominantly focus on holistic naturalness, leaving fine-grained paralinguistic distinctions underexplored. We introduce ParaPairAudioBench, a pairwise benchmark of 5,175 audio pairs across five paralinguistic dimensions: Style, Rate, Emphasis, Age, and Gender. Our experiments show that current LALM judges still lag behind human judgments by 32%p on average and exhibit severe calibration failures, particularly in Tie cases where the correct decision is to abstain. To further analyze lexical versus acoustic reliance, the benchmark includes both same-transcript and cross-transcript conditions. ParaPairAudioBench enables multi-dimensional, calibration-aware assessment of the reliability of LALM-as-a-Judge for paralinguistic speech evaluation.

VieSpeaker: A Large-Scale Vietnamese Speaker Recognition Dataset Beyond Visual Dependency

Speaker recognition has advanced rapidly with large-scale training datasets, yet Vietnamese remains under-resourced, with existing corpora limited in scale and acoustic diversity. Most large-scale datasets rely on facial cues to link speech with speaker identities, restricting data collection to recordings where speakers appear on camera. We propose a face-independent dataset construction pipeline and introduce VieSpeaker, a large-scale Vietnamese speaker recognition dataset. Our approach leverages textual metadata and large language model reasoning to infer speaker identities from transcripts and contextual information. VieSpeaker contains approximately 902 hours of speech from 4,715 speakers. Experiments show that models trained on VieSpeaker achieve improved robustness and generalization compared to existing Vietnamese datasets. This work demonstrates the feasibility of face-independent dataset construction and provides a new direction for building large-scale speech resources.

The Watermark Shortcut: How Provenance Marking Sabotages Audio Deepfake Detection

Provenance watermarking is increasingly treated as a safeguard for synthetic speech, whether built directly into speech-generation models such as Chatterbox, provided through dedicated techniques such as AudioSeal, or deployed by commercial platforms such as ElevenLabs. We identify a previously uncharacterized liability: when synthetic speech is watermarked and human speech is not, detectors trained alongside latch onto the watermark as a spurious "watermark => fake" shortcut. This single feature yields three coupled failures: generalization degradation (model performance deteriorates on unseen data), strip-to-evade (a watermarked fake escapes once unwatermarked), and mark-to-frame (watermarking a real voice flags it as fake). In a controlled white-box experiment, a watermark-trained detector shows all three (for example, mark-to-frame lifts Equal Error Rate from 16% to 75%). In a black-box test of a commercial API, we show that adding a watermark to real speech disguises it as fake. However, this shortcut is fixable: retraining with the watermark on both classes decorrelates it and restores clean behavior. We release experiment data as a paired clean-versus-watermarked corpus (WASP).

Don't Listen to Me: A Lightweight, Low-Latency Model for Own-Voice Cancellation in Far-Field Speech Enhancement

We introduce own-voice cancellation (OVC): removing a target (enrolled) speaker from a noisy multi-speaker mixture while preserving any remaining speech. Framed as the complement of target speaker extraction, OVC addresses latency-induced own-voice artifacts that arise when a far-field device streams enhanced audio back to the user, as the round-trip time easily exceeds the perceptual threshold for own-voice distortion. We condition a time-domain model with only 2 ms algorithmic latency on a short enrollment utterance and benchmark TD-SpeakerBeam alongside a lighter Mamba-MinGRU masker built from Mamba blocks with MinGRU temporal mixing. Replacing the ConvTasNet-based auxiliary network with a linear RNN encoder improves both signal-to-distortion ratio and predicted MOS while reducing compute. Results establish OVC as a practical, low-latency enhancement objective for far-field denoising.

On the Effect of Segmentation Width and Cluster Size on Speech Resynthesis and Continuation in Generative Spoken Language Models

Generative Spoken Language Modeling (GSLM) enables text-free speech modeling by training language models (LMs) using discrete speech representations instead of textual transcription. In this paper, we investigate the performance of GSLM on speech synthesis and continuation using discrete speech representations with varying bitrates. We segment speech representations with fixed widths and train K-means models in multiple cluster sizes, resulting in various bitrate settings. We demonstrate that intelligible and natural speech can be synthesized at lower bitrate settings than the baseline. Furthermore, speech continuation quality remains stable at lower bitrates across multiple metrics, suggesting that the conventional GSLM setting may be redundant for effective speech generation. Although LLM-based metrics show higher correlation with human subjective score than conventional metrics, it remains low, highlighting the need for more stable automatic evaluation methods.

Acoustic Landmark Detector based on Conformer and HuBERT

Acoustic landmarks (abrupt acoustic changes tied to speech events) offer a linguistically grounded representation for speech analysis. We study automatic landmark detection with Conformer models, evaluating 14 configurations spanning architecture, loss, label representation, feature extractor, and data conditions on 1 839 manually annotated utterances with eight landmark types. We propose Gaussian soft labels with per-class temporal spread (sigma=10-20 ms), improving F1-at-20 ms by 7.0% absolute vs. hard labels by modeling annotation variability. Frozen HuBERT features perform best without fine-tuning (F1-at-20 ms=0.77). Stops and fricatives are reliable (F1>0.80), while vowels remain challenging (F1 approx 0.55). On our corpus, our system reaches a 13.8% Landmark Error Rate (LER). This is not directly comparable to AutoLandmark (31.3%) or SpeechMark (56.5%), evaluated on a different corpus and metric. Per-class trends show detectability increases with event abruptness, consistent with Stevens' theory.

An Acoustic Landmark Database of the English Lexicon via Articulatory Synthesis

Acoustic landmark theory treats speech as organized around the acoustic consequences of articulatory gestures that shape the vocal tract and airflow. Progress is limited by the scarcity of large, unambiguously annotated landmark datasets. We invert the problem by generating speech from landmark patterns. Using the Pink Trombone physical vocal-tract synthesizer, we produce an English lexicon for two adult configurations (male, female). With direct control of gestures, we place landmark labels algorithmically at the exact times of their physical events (e.g., oral closures/releases). The corpus contains $>$200,000 synthesized words, rendered for both configurations with time-aligned annotations; intelligibility is measured with STOI. We leverage it for statistics across the lexicon from an articulatory-event view, reporting landmark frequencies and dominant cue patterns, and enabling quantitative studies plus training/benchmarking of automatic landmark detectors.