Private kNN-VC: Interpretable Anonymization of Converted Speech

Speaker anonymization seeks to conceal a speaker's identity while preserving the utility of their speech. The achieved privacy is commonly evaluated with a speaker recognition model trained on anonymized speech. Although this represents a strong attack, it is unclear which aspects of speech are exploited to identify the speakers. Our research sets out to unveil these aspects. It starts with kNN-VC, a powerful voice conversion model that performs poorly as an anonymization system, presumably because of prosody leakage. To test this hypothesis, we extend kNN-VC with two interpretable components that anonymize the duration and variation of phones. These components increase privacy significantly, proving that the studied prosodic factors encode speaker identity and are exploited by the privacy attack. Additionally, we show that changes in the target selection algorithm considerably influence the outcome of the privacy attack.

Dual Precision Quantization for Efficient and Accurate Deep Neural Networks Inference

Deep neural networks have achieved state-of-the-art results in a wide range of applications, from natural language processing and computer vision to speech recognition. However, as tasks become increasingly complex, model sizes continue to grow, posing challenges in latency and memory efficiency. To meet these constraints, post-training quantization has emerged as a promising solution. In this paper, we propose a novel hardware-efficient quantization and inference scheme that exploits hardware advantages with minimal accuracy degradation. Specifically, we introduce a W4A8 scheme, where weights are quantized and stored using 4-bit integer precision, and inference computations are performed using 8-bit floating-point arithmetic, demonstrating significant speedups and improved memory utilization compared to 16-bit operations, applicable on various modern accelerators. To mitigate accuracy loss, we develop a novel quantization algorithm, dubbed Dual Precision Quantization (DPQ), that leverages the unique structure of our scheme without introducing additional inference overhead. Experimental results demonstrate improved performance (i.e., increased throughput) while maintaining tolerable accuracy degradation relative to the full-precision model.

Vox-Profile: A Speech Foundation Model Benchmark for Characterizing Diverse Speaker and Speech Traits

We introduce Vox-Profile, a comprehensive benchmark to characterize rich speaker and speech traits using speech foundation models. Unlike existing works that focus on a single dimension of speaker traits, Vox-Profile provides holistic and multi-dimensional profiles that reflect both static speaker traits (e.g., age, sex, accent) and dynamic speech properties (e.g., emotion, speech flow). This benchmark is grounded in speech science and linguistics, developed with domain experts to accurately index speaker and speech characteristics. We report benchmark experiments using over 15 publicly available speech datasets and several widely used speech foundation models that target various static and dynamic speaker and speech properties. In addition to benchmark experiments, we showcase several downstream applications supported by Vox-Profile. First, we show that Vox-Profile can augment existing speech recognition datasets to analyze ASR performance variability. Vox-Profile is also used as a tool to evaluate the performance of speech generation systems. Finally, we assess the quality of our automated profiles through comparison with human evaluation and show convergent validity. Vox-Profile is publicly available at: https://github.com/tiantiaf0627/vox-profile-release.

Survey of End-to-End Multi-Speaker Automatic Speech Recognition for Monaural Audio

Monaural multi-speaker automatic speech recognition (ASR) remains challenging due to data scarcity and the intrinsic difficulty of recognizing and attributing words to individual speakers, particularly in overlapping speech. Recent advances have driven the shift from cascade systems to end-to-end (E2E) architectures, which reduce error propagation and better exploit the synergy between speech content and speaker identity. Despite rapid progress in E2E multi-speaker ASR, the field lacks a comprehensive review of recent developments. This survey provides a systematic taxonomy of E2E neural approaches for multi-speaker ASR, highlighting recent advances and comparative analysis. Specifically, we analyze: (1) architectural paradigms (SIMO vs.~SISO) for pre-segmented audio, analyzing their distinct characteristics and trade-offs; (2) recent architectural and algorithmic improvements based on these two paradigms; (3) extensions to long-form speech, including segmentation strategy and speaker-consistent hypothesis stitching. Further, we (4) evaluate and compare methods across standard benchmarks. We conclude with a discussion of open challenges and future research directions towards building robust and scalable multi-speaker ASR.

Automatic Speech Recognition for African Low-Resource Languages: Challenges and Future Directions

Automatic Speech Recognition (ASR) technologies have transformed human-computer interaction; however, low-resource languages in Africa remain significantly underrepresented in both research and practical applications. This study investigates the major challenges hindering the development of ASR systems for these languages, which include data scarcity, linguistic complexity, limited computational resources, acoustic variability, and ethical concerns surrounding bias and privacy. The primary goal is to critically analyze these barriers and identify practical, inclusive strategies to advance ASR technologies within the African context. Recent advances and case studies emphasize promising strategies such as community-driven data collection, self-supervised and multilingual learning, lightweight model architectures, and techniques that prioritize privacy. Evidence from pilot projects involving various African languages showcases the feasibility and impact of customized solutions, which encompass morpheme-based modeling and domain-specific ASR applications in sectors like healthcare and education. The findings highlight the importance of interdisciplinary collaboration and sustained investment to tackle the distinct linguistic and infrastructural challenges faced by the continent. This study offers a progressive roadmap for creating ethical, efficient, and inclusive ASR systems that not only safeguard linguistic diversity but also improve digital accessibility and promote socioeconomic participation for speakers of African languages.

Calm-Whisper: Reduce Whisper Hallucination On Non-Speech By Calming Crazy Heads Down

OpenAI's Whisper has achieved significant success in Automatic Speech Recognition. However, it has consistently been found to exhibit hallucination issues, particularly in non-speech segments, which limits its broader application in complex industrial settings. In this paper, we introduce a novel method to reduce Whisper's hallucination on non-speech segments without using any pre- or post-possessing techniques. Specifically, we benchmark the contribution of each self-attentional head in the Whisper-large-v3 decoder to the hallucination problem by performing a head-wise mask. Our findings reveal that only 3 of the 20 heads account for over 75% of the hallucinations on the UrbanSound dataset. We then fine-tune these three crazy heads using a collection of non-speech data. The results show that our best fine-tuned model, namely Calm-Whisper, achieves over 80% reduction in non-speech hallucination with only less than 0.1% WER degradation on LibriSpeech test-clean and test-other.

KIT's Offline Speech Translation and Instruction Following Submission for IWSLT 2025

The scope of the International Workshop on Spoken Language Translation (IWSLT) has recently broadened beyond traditional Speech Translation (ST) to encompass a wider array of tasks, including Speech Question Answering and Summarization. This shift is partly driven by the growing capabilities of modern systems, particularly with the success of Large Language Models (LLMs). In this paper, we present the Karlsruhe Institute of Technology's submissions for the Offline ST and Instruction Following (IF) tracks, where we leverage LLMs to enhance performance across all tasks. For the Offline ST track, we propose a pipeline that employs multiple automatic speech recognition systems, whose outputs are fused using an LLM with document-level context. This is followed by a two-step translation process, incorporating additional refinement step to improve translation quality. For the IF track, we develop an end-to-end model that integrates a speech encoder with an LLM to perform a wide range of instruction-following tasks. We complement it with a final document-level refinement stage to further enhance output quality by using contextual information.

FeatureSense: Protecting Speaker Attributes in Always-On Audio Sensing System

Audio is a rich sensing modality that is useful for a variety of human activity recognition tasks. However, the ubiquitous nature of smartphones and smart speakers with always-on microphones has led to numerous privacy concerns and a lack of trust in deploying these audio-based sensing systems. This paper addresses this critical challenge of preserving user privacy when using audio for sensing applications while maintaining utility. While prior work focuses primarily on protecting recoverable speech content, we show that sensitive speaker-specific attributes such as age and gender can still be inferred after masking speech and propose a comprehensive privacy evaluation framework to assess this speaker attribute leakage. We design and implement FeatureSense, an open-source library that provides a set of generalizable privacy-aware audio features that can be used for wide range of sensing applications. We present an adaptive task-specific feature selection algorithm that optimizes the privacy-utility-cost trade-off based on the application requirements. Through our extensive evaluation, we demonstrate the high utility of FeatureSense across a diverse set of sensing tasks. Our system outperforms existing privacy techniques by 60.6% in preserving user-specific privacy. This work provides a foundational framework for ensuring trust in audio sensing by enabling effective privacy-aware audio classification systems.

Cross-modal Knowledge Transfer Learning as Graph Matching Based on Optimal Transport for ASR

Transferring linguistic knowledge from a pretrained language model (PLM) to acoustic feature learning has proven effective in enhancing end-to-end automatic speech recognition (E2E-ASR). However, aligning representations between linguistic and acoustic modalities remains a challenge due to inherent modality gaps. Optimal transport (OT) has shown promise in mitigating these gaps by minimizing the Wasserstein distance (WD) between linguistic and acoustic feature distributions. However, previous OT-based methods overlook structural relationships, treating feature vectors as unordered sets. To address this, we propose Graph Matching Optimal Transport (GM-OT), which models linguistic and acoustic sequences as structured graphs. Nodes represent feature embeddings, while edges capture temporal and sequential relationships. GM-OT minimizes both WD (between nodes) and Gromov-Wasserstein distance (GWD) (between edges), leading to a fused Gromov-Wasserstein distance (FGWD) formulation. This enables structured alignment and more efficient knowledge transfer compared to existing OT-based approaches. Theoretical analysis further shows that prior OT-based methods in linguistic knowledge transfer can be viewed as a special case within our GM-OT framework. We evaluate GM-OT on Mandarin ASR using a CTC-based E2E-ASR system with a PLM for knowledge transfer. Experimental results demonstrate significant performance gains over state-of-the-art models, validating the effectiveness of our approach.

Pretraining Multi-Speaker Identification for Neural Speaker Diarization

End-to-end speaker diarization enables accurate overlap-aware diarization by jointly estimating multiple speakers' speech activities in parallel. This approach is data-hungry, requiring a large amount of labeled conversational data, which cannot be fully obtained from real datasets alone. To address this issue, large-scale simulated data is often used for pretraining, but it requires enormous storage and I/O capacity, and simulating data that closely resembles real conversations remains challenging. In this paper, we propose pretraining a model to identify multiple speakers from an input fully overlapped mixture as an alternative to pretraining a diarization model. This method eliminates the need to prepare a large-scale simulated dataset while leveraging large-scale speaker recognition datasets for training. Through comprehensive experiments, we demonstrate that the proposed method enables a highly accurate yet lightweight local diarization model without simulated conversational data.