Sanjeev Khudanpur

Samuele Cornell, Matthew Wiesner, Shinji Watanabe et al. · cmu

The CHiME challenges have played a significant role in the development and evaluation of robust automatic speech recognition (ASR) systems. We introduce the CHiME-7 distant ASR (DASR) task, within the 7th CHiME challenge. This task comprises joint ASR and diarization in far-field settings with multiple, and possibly heterogeneous, recording devices. Different from previous challenges, we evaluate systems on 3 diverse scenarios: CHiME-6, DiPCo, and Mixer 6. The goal is for participants to devise a single system that can generalize across different array geometries and use cases with no a-priori information. Another departure from earlier CHiME iterations is that participants are allowed to use open-source pre-trained models and datasets. In this paper, we describe the challenge design, motivation, and fundamental research questions in detail. We also present the baseline system, which is fully array-topology agnostic and features multi-channel diarization, channel selection, guided source separation and a robust ASR model that leverages self-supervised speech representations (SSLR).

Cihan Xiao, Yiwen Shao, Chenxing Li et al.

Audio and omni-modal large language models exhibit impressive cross-modal reasoning capabilities. However, applying standard reinforcement learning post-training algorithms to these models exposes a critical structural vulnerability: methods like GRPO apply uniform policy gradients across all tokens, ignoring their unequal dependence on the non-text source modality. This exacerbates late-stage modality collapse during extended chain-of-thought generation, where models progressively abandon the primary source signal in favor of compressed textual priors, leading to confident but ungrounded hallucinations. To address this, we introduce Modality-Aware Policy Optimization (MAPO), a novel dual-branch reinforcement learning framework. First, MAPO dynamically concentrates the policy gradient on modality-critical tokens using a modality relevance mask, which is derived from the cross-modal differential entropy between an audio-ablated reference and the multimodal policy. Second, it integrates an auxiliary attention loss branch that applies a targeted, temporally scaled penalty to the model's internal attention distributions. This ensures the model actively sustains cross-modal grounding deep into the reasoning trace. Evaluations on complex audio reasoning benchmarks demonstrate that MAPO substantially improves long-horizon reasoning fidelity and multimodal instruction following, achieving highly competitive performance and setting new state-of-the-art results on several key benchmarks among open-weight models. By relying strictly on native statistical signals rather than domain-specific inductive biases, MAPO offers a promising foundation for mitigating epistemic collapse across diverse multimodal systems.

Dongji Gao, Jiatong Shi, Shun-Po Chuang et al.

This paper describes the ESPnet Unsupervised ASR Open-source Toolkit (EURO), an end-to-end open-source toolkit for unsupervised automatic speech recognition (UASR). EURO adopts the state-of-the-art UASR learning method introduced by the Wav2vec-U, originally implemented at FAIRSEQ, which leverages self-supervised speech representations and adversarial training. In addition to wav2vec2, EURO extends the functionality and promotes reproducibility for UASR tasks by integrating S3PRL and k2, resulting in flexible frontends from 27 self-supervised models and various graph-based decoding strategies. EURO is implemented in ESPnet and follows its unified pipeline to provide UASR recipes with a complete setup. This improves the pipeline's efficiency and allows EURO to be easily applied to existing datasets in ESPnet. Extensive experiments on three mainstream self-supervised models demonstrate the toolkit's effectiveness and achieve state-of-the-art UASR performance on TIMIT and LibriSpeech datasets. EURO will be publicly available at https://github.com/espnet/espnet, aiming to promote this exciting and emerging research area based on UASR through open-source activity.

Amir Hussein, Brian Yan, Antonios Anastasopoulos et al. · cmu

Incorporating longer context has been shown to benefit machine translation, but the inclusion of context in end-to-end speech translation (E2E-ST) remains under-studied. To bridge this gap, we introduce target language context in E2E-ST, enhancing coherence and overcoming memory constraints of extended audio segments. Additionally, we propose context dropout to ensure robustness to the absence of context, and further improve performance by adding speaker information. Our proposed contextual E2E-ST outperforms the isolated utterance-based E2E-ST approach. Lastly, we demonstrate that in conversational speech, contextual information primarily contributes to capturing context style, as well as resolving anaphora and named entities.

Yining Lu, Dixuan Wang, Tianjian Li et al.

As LLMs become increasingly prevalent, it is interesting to consider how ``creative'' these models can be. From cognitive science, creativity consists of at least two key characteristics: \emph{convergent} thinking (purposefulness to achieve a given goal) and \emph{divergent} thinking (adaptability to explore new environments or constraints) \citep{runco2003critical}. In this work, we introduce a framework for quantifying LLM creativity that incorporates the two design ingredients: (1) We introduce DENIAL PROMPTING which pushes LLMs to develop more creative solutions to a given problem by incrementally imposing new constraints on the previous solution, compelling LLMs to adopt new strategies. (2) We define NEOGAUGE, a metric that quantifies both convergent and divergent thinking in the generated creative responses by LLMs. We test the proposed framework on Codeforces problems, which serve as both a natural dataset for coding tasks and a collection of prior human solutions. We quantify NEOGAUGE for various proprietary and open-source models and find that even the most creative model, GPT-4, still falls short of demonstrating human-like creativity. We also experiment with advanced reasoning strategies (MCTS, self-correction, etc.) and observe no significant improvement in creativity. As a by-product of our analysis, we release NEOCODER dataset for reproducing our results on future models.

Dongji Gao, Matthew Wiesner, Hainan Xu et al.

This paper presents a novel algorithm for building an automatic speech recognition (ASR) model with imperfect training data. Imperfectly transcribed speech is a prevalent issue in human-annotated speech corpora, which degrades the performance of ASR models. To address this problem, we propose Bypass Temporal Classification (BTC) as an expansion of the Connectionist Temporal Classification (CTC) criterion. BTC explicitly encodes the uncertainties associated with transcripts during training. This is accomplished by enhancing the flexibility of the training graph, which is implemented as a weighted finite-state transducer (WFST) composition. The proposed algorithm improves the robustness and accuracy of ASR systems, particularly when working with imprecisely transcribed speech corpora. Our implementation will be open-sourced.

Dongji Gao, Hainan Xu, Desh Raj et al.

Training automatic speech recognition (ASR) systems requires large amounts of well-curated paired data. However, human annotators usually perform "non-verbatim" transcription, which can result in poorly trained models. In this paper, we propose Omni-temporal Classification (OTC), a novel training criterion that explicitly incorporates label uncertainties originating from such weak supervision. This allows the model to effectively learn speech-text alignments while accommodating errors present in the training transcripts. OTC extends the conventional CTC objective for imperfect transcripts by leveraging weighted finite state transducers. Through experiments conducted on the LibriSpeech and LibriVox datasets, we demonstrate that training ASR models with OTC avoids performance degradation even with transcripts containing up to 70% errors, a scenario where CTC models fail completely. Our implementation is available at https://github.com/k2-fsa/icefall.

Matthew Wiesner, Samuele Cornell, Alexander Polok et al. · cmu

We propose to model parallel streams of data, such as overlapped speech, using shuffles. Specifically, this paper shows how the shuffle product and partial order finite-state automata (FSAs) can be used for alignment and speaker-attributed transcription of overlapped speech. We train using the total score on these FSAs as a loss function, marginalizing over all possible serializations of overlapping sequences at subword, word, and phrase levels. To reduce graph size, we impose temporal constraints by constructing partial order FSAs. We address speaker attribution by modeling (token, speaker) tuples directly. Viterbi alignment through the shuffle product FSA directly enables one-pass alignment. We evaluate performance on synthetic LibriSpeech overlaps. To our knowledge, this is the first algorithm that enables single-pass alignment of multi-talker recordings. All algorithms are implemented using k2 / Icefall.

Amir Hussein, Dorsa Zeinali, Ondřej Klejch et al.

Designing effective automatic speech recognition (ASR) systems for Code-Switching (CS) often depends on the availability of the transcribed CS resources. To address data scarcity, this paper introduces Speech Collage, a method that synthesizes CS data from monolingual corpora by splicing audio segments. We further improve the smoothness quality of audio generation using an overlap-add approach. We investigate the impact of generated data on speech recognition in two scenarios: using in-domain CS text and a zero-shot approach with synthesized CS text. Empirical results highlight up to 34.4% and 16.2% relative reductions in Mixed-Error Rate and Word-Error Rate for in-domain and zero-shot scenarios, respectively. Lastly, we demonstrate that CS augmentation bolsters the model's code-switching inclination and reduces its monolingual bias.

Henry Li Xinyuan, Zexin Cai, Ashi Garg et al.

We present a number of systems for the Voice Privacy Challenge, including voice conversion based systems such as the kNN-VC method and the WavLM voice Conversion method, and text-to-speech (TTS) based systems including Whisper-VITS. We found that while voice conversion systems better preserve emotional content, they struggle to conceal speaker identity in semi-white-box attack scenarios; conversely, TTS methods perform better at anonymization and worse at emotion preservation. Finally, we propose a random admixture system which seeks to balance out the strengths and weaknesses of the two category of systems, achieving a strong EER of over 40% while maintaining UAR at a respectable 47%.

Ruizhe Huang, Mahsa Yarmohammadi, Sanjeev Khudanpur et al.

Existing research suggests that automatic speech recognition (ASR) models can benefit from additional contexts (e.g., contact lists, user specified vocabulary). Rare words and named entities can be better recognized with contexts. In this work, we propose two simple yet effective techniques to improve context-aware ASR models. First, we inject contexts into the encoders at an early stage instead of merely at their last layers. Second, to enforce the model to leverage the contexts during training, we perturb the reference transcription with alternative spellings so that the model learns to rely on the contexts to make correct predictions. On LibriSpeech, our techniques together reduce the rare word error rate by 60% and 25% relatively compared to no biasing and shallow fusion, making the new state-of-the-art performance. On SPGISpeech and a real-world dataset ConEC, our techniques also yield good improvements over the baselines.

Zexin Cai, Henry Li Xinyuan, Ashi Garg et al.

Advances in speech technology now allow unprecedented access to personally identifiable information through speech. To protect such information, the differential privacy field has explored ways to anonymize speech while preserving its utility, including linguistic and paralinguistic aspects. However, anonymizing speech while maintaining emotional state remains challenging. We explore this problem in the context of the VoicePrivacy 2024 challenge. Specifically, we developed various speaker anonymization pipelines and find that approaches either excel at anonymization or preserving emotion state, but not both simultaneously. Achieving both would require an in-domain emotion recognizer. Additionally, we found that it is feasible to train a semi-effective speaker verification system using only emotion representations, demonstrating the challenge of separating these two modalities.

Hexin Liu, Leibny Paola Garcia Perera, Andy W. H. Khong et al.

In this paper, we propose to employ a dual-mode framework on the x-vector self-attention (XSA-LID) model with knowledge distillation (KD) to enhance its language identification (LID) performance for both long and short utterances. The dual-mode XSA-LID model is trained by jointly optimizing both the full and short modes with their respective inputs being the full-length speech and its short clip extracted by a specific Boolean mask, and KD is applied to further boost the performance on short utterances. In addition, we investigate the impact of clip-wise linguistic variability and lexical integrity for LID by analyzing the variation of LID performance in terms of the lengths and positions of the mimicked speech clips. We evaluated our approach on the MLS14 data from the NIST 2017 LRE. With the 3~s random-location Boolean mask, our proposed method achieved 19.23%, 21.52% and 8.37% relative improvement in average cost compared with the XSA-LID model on 3s, 10s, and 30s speech, respectively.

Cihan Xiao, Henry Li Xinyuan, Jinyi Yang et al.

We introduce HK-LegiCoST, a new three-way parallel corpus of Cantonese-English translations, containing 600+ hours of Cantonese audio, its standard traditional Chinese transcript, and English translation, segmented and aligned at the sentence level. We describe the notable challenges in corpus preparation: segmentation, alignment of long audio recordings, and sentence-level alignment with non-verbatim transcripts. Such transcripts make the corpus suitable for speech translation research when there are significant differences between the spoken and written forms of the source language. Due to its large size, we are able to demonstrate competitive speech translation baselines on HK-LegiCoST and extend them to promising cross-corpus results on the FLEURS Cantonese subset. These results deliver insights into speech recognition and translation research in languages for which non-verbatim or ``noisy'' transcription is common due to various factors, including vernacular and dialectal speech.

Henry Li Xinyuan, Sonal Joshi, Thomas Thebaud et al.

Poisoning backdoor attacks involve an adversary manipulating the training data to induce certain behaviors in the victim model by inserting a trigger in the signal at inference time. We adapted clean label backdoor (CLBD)-data poisoning attacks, which do not modify the training labels, on state-of-the-art speech recognition models that support/perform a Spoken Language Understanding task, achieving 99.8% attack success rate by poisoning 10% of the training data. We analyzed how varying the signal-strength of the poison, percent of samples poisoned, and choice of trigger impact the attack. We also found that CLBD attacks are most successful when applied to training samples that are inherently hard for a proxy model. Using this strategy, we achieved an attack success rate of 99.3% by poisoning a meager 1.5% of the training data. Finally, we applied two previously developed defenses against gradient-based attacks, and found that they attain mixed success against poisoning.

Lin Zhang, Thomas Thebaud, Zexin Cai et al.

Large language models (LLMs), trained on large-scale text, have recently attracted significant attention for their strong performance across many tasks. Motivated by this, we investigate whether a text-trained LLM can help localize fake words in partially fake speech, where only specific words within a speech are edited. We build a speech LLM to perform fake word localization via next token prediction. Experiments and analyses on AV-Deepfake1M and PartialEdit indicates that the model frequently leverages editing-style pattern learned from the training data, particularly word-level polarity substitutions for those two databases we discussed, as cues for localizing fake words. Although such particular patterns provide useful information in an in-domain scenario, how to avoid over-reliance on such particular pattern and improve generalization to unseen editing styles remains an open question.

Kai Tan, Lin Zhang, Ruiteng Zhang et al.

Spoofing-robust automatic speaker verification (SASV) aims to integrate automatic speaker verification (ASV) and countermeasure (CM). A popular solution is fusion of independent ASV and CM scores. To better modeling SASV, some frameworks integrate ASV and CM within a single network. However, these solutions are typically bi-encoder based, offer limited interpretability, and cannot be readily adapted to new evaluation parameters without retraining. Based on this, we propose a unified end-to-end framework via a three-class formulation that enables log-likelihood ratio (LLR) inference from class logits for a more interpretable decision pipeline. Experiments show comparable performance to existing methods on ASVSpoof5 and better results on SpoofCeleb. The visualization and analysis also prove that the three-class reformulation provides more interpretability.

Dongji Gao, Chenda Liao, Changliang Liu et al.

The Recurrent Neural Network-Transducer (RNN-T) is widely adopted in end-to-end (E2E) automatic speech recognition (ASR) tasks but depends heavily on large-scale, high-quality annotated data, which are often costly and difficult to obtain. To mitigate this reliance, we propose a Weakly Supervised Transducer (WST), which integrates a flexible training graph designed to robustly handle errors in the transcripts without requiring additional confidence estimation or auxiliary pre-trained models. Empirical evaluations on synthetic and industrial datasets reveal that WST effectively maintains performance even with transcription error rates of up to 70%, consistently outperforming existing Connectionist Temporal Classification (CTC)-based weakly supervised approaches, such as Bypass Temporal Classification (BTC) and Omni-Temporal Classification (OTC). These results demonstrate the practical utility and robustness of WST in realistic ASR settings. The implementation will be publicly available.

Brian Yan, Injy Hamed, Shuichiro Shimizu et al. · cmu

We present CS-FLEURS, a new dataset for developing and evaluating code-switched speech recognition and translation systems beyond high-resourced languages. CS-FLEURS consists of 4 test sets which cover in total 113 unique code-switched language pairs across 52 languages: 1) a 14 X-English language pair set with real voices reading synthetically generated code-switched sentences, 2) a 16 X-English language pair set with generative text-to-speech 3) a 60 {Arabic, Mandarin, Hindi, Spanish}-X language pair set with the generative text-to-speech, and 4) a 45 X-English lower-resourced language pair test set with concatenative text-to-speech. Besides the four test sets, CS-FLEURS also provides a training set with 128 hours of generative text-to-speech data across 16 X-English language pairs. Our hope is that CS-FLEURS helps to broaden the scope of future code-switched speech research. Dataset link: https://huggingface.co/datasets/byan/cs-fleurs.

Desh Raj, Sanjeev Khudanpur

We recently proposed DOVER-Lap, a method for combining overlap-aware speaker diarization system outputs. DOVER-Lap improved upon its predecessor DOVER by using a label mapping method based on globally-informed greedy search. In this paper, we analyze this label mapping in the framework of a maximum orthogonal graph partitioning problem, and present three inferences. First, we show that DOVER-Lap label mapping is exponential in the input size, which poses a challenge when combining a large number of hypotheses. We then revisit the DOVER label mapping algorithm and propose a modification which performs similar to DOVER-Lap while being computationally tractable. We also derive an approximation bound for the algorithm in terms of the maximum number of hypotheses speakers. Finally, we describe a randomized local search algorithm which provides a near-optimal $(1-ε)$-approximate solution to the problem with high probability. We empirically demonstrate the effectiveness of our methods on the AMI meeting corpus. Our code is publicly available: https://github.com/desh2608/dover-lap.

Shinji Watanabe, Michael Mandel, Jon Barker et al.

Following the success of the 1st, 2nd, 3rd, 4th and 5th CHiME challenges we organize the 6th CHiME Speech Separation and Recognition Challenge (CHiME-6). The new challenge revisits the previous CHiME-5 challenge and further considers the problem of distant multi-microphone conversational speech diarization and recognition in everyday home environments. Speech material is the same as the previous CHiME-5 recordings except for accurate array synchronization. The material was elicited using a dinner party scenario with efforts taken to capture data that is representative of natural conversational speech. This paper provides a baseline description of the CHiME-6 challenge for both segmented multispeaker speech recognition (Track 1) and unsegmented multispeaker speech recognition (Track 2). Of note, Track 2 is the first challenge activity in the community to tackle an unsegmented multispeaker speech recognition scenario with a complete set of reproducible open source baselines providing speech enhancement, speaker diarization, and speech recognition modules.

Yiming Wang, Tongfei Chen, Hainan Xu et al.

We present Espresso, an open-source, modular, extensible end-to-end neural automatic speech recognition (ASR) toolkit based on the deep learning library PyTorch and the popular neural machine translation toolkit fairseq. Espresso supports distributed training across GPUs and computing nodes, and features various decoding approaches commonly employed in ASR, including look-ahead word-based language model fusion, for which a fast, parallelized decoder is implemented. Espresso achieves state-of-the-art ASR performance on the WSJ, LibriSpeech, and Switchboard data sets among other end-to-end systems without data augmentation, and is 4--11x faster for decoding than similar systems (e.g. ESPnet).

Sandra Arcos-Holzinger, Sarah M. Erfani, James Bailey et al.

Self-supervised speech models (S3Ms) achieve strong downstream performance, yet their learned representations remain poorly understood under natural and adversarial perturbations. Prior studies rely on representation similarity or global dimensionality, offering limited visibility into local geometric changes. We ask: how do perturbations deform local geometry, and do these shifts track downstream automatic speech recognition (ASR) degradation? To address this, we present GRIDS, a framework using Local Intrinsic Dimensionality (LID) across layer-wise representations in WavLM and wav2vec 2.0. We find that LID increases for all low signal-to noise ratio (SNR) perturbations and diverges at high SNR: benign noise converges toward the clean profile, while adversarial inputs retain early-layer LID elevation. We show LID elevation co-occurs with increased WER, and that layer-wise LID features enable anomaly detection (AUROC 0.78-1.00), opening the door to transcript-free monitoring in S3Ms.

Nathaniel R. Robinson, Raj Dabre, Ammon Shurtz et al.

A majority of language technologies are tailored for a small number of high-resource languages, while relatively many low-resource languages are neglected. One such group, Creole languages, have long been marginalized in academic study, though their speakers could benefit from machine translation (MT). These languages are predominantly used in much of Latin America, Africa and the Caribbean. We present the largest cumulative dataset to date for Creole language MT, including 14.5M unique Creole sentences with parallel translations -- 11.6M of which we release publicly, and the largest bitexts gathered to date for 41 languages -- the first ever for 21. In addition, we provide MT models supporting all 41 Creole languages in 172 translation directions. Given our diverse dataset, we produce a model for Creole language MT exposed to more genre diversity than ever before, which outperforms a genre-specific Creole MT model on its own benchmark for 26 of 34 translation directions.

Ruizhe Huang, Xiaohui Zhang, Zhaoheng Ni et al.

Connectionist temporal classification (CTC) models are known to have peaky output distributions. Such behavior is not a problem for automatic speech recognition (ASR), but it can cause inaccurate forced alignments (FA), especially at finer granularity, e.g., phoneme level. This paper aims at alleviating the peaky behavior for CTC and improve its suitability for forced alignment generation, by leveraging label priors, so that the scores of alignment paths containing fewer blanks are boosted and maximized during training. As a result, our CTC model produces less peaky posteriors and is able to more accurately predict the offset of the tokens besides their onset. It outperforms the standard CTC model and a heuristics-based approach for obtaining CTC's token offset timestamps by 12-40% in phoneme and word boundary errors (PBE and WBE) measured on the Buckeye and TIMIT data. Compared with the most widely used FA toolkit Montreal Forced Aligner (MFA), our method performs similarly on PBE/WBE on Buckeye, yet falls behind MFA on TIMIT. Nevertheless, our method has a much simpler training pipeline and better runtime efficiency. Our training recipe and pretrained model are released in TorchAudio.

Zexin Cai, Henry Li Xinyuan, Ashi Garg et al.

Most current zero-shot voice conversion methods rely on externally supervised components, particularly speaker encoders, for training. To explore alternatives that eliminate this dependency, this paper introduces GenVC, a novel framework that disentangles speaker identity and linguistic content from speech signals in a self-supervised manner. GenVC leverages speech tokenizers and an autoregressive, Transformer-based language model as its backbone for speech generation. This design supports large-scale training while enhancing both source speaker privacy protection and target speaker cloning fidelity. Experimental results demonstrate that GenVC achieves notably higher speaker similarity, with naturalness on par with leading zero-shot approaches. Moreover, due to its autoregressive formulation, GenVC introduces flexibility in temporal alignment, reducing the preservation of source prosody and speaker-specific traits, and making it highly effective for voice anonymization.

Alexander Polok, Dominik Klement, Samuele Cornell et al.

Speaker-attributed automatic speech recognition (ASR) in multi-speaker environments remains a major challenge. While some approaches achieve strong performance when fine-tuned on specific domains, few systems generalize well across out-of-domain datasets. Our prior work, Diarization-Conditioned Whisper (DiCoW), leverages speaker diarization outputs as conditioning information and, with minimal fine-tuning, demonstrated strong multilingual and multi-domain performance. In this paper, we address a key limitation of DiCoW: ambiguity in Silence-Target-Non-target-Overlap (STNO) masks, where two or more fully overlapping speakers may have nearly identical conditioning despite differing transcriptions. We introduce SE-DiCoW (Self-Enrolled Diarization-Conditioned Whisper), which uses diarization output to locate an enrollment segment anywhere in the conversation where the target speaker is most active. This enrollment segment is used as fixed conditioning via cross-attention at each encoder layer. We further refine DiCoW with improved data segmentation, model initialization, and augmentation. Together, these advances yield substantial gains: SE-DiCoW reduces macro-averaged tcpWER by 52.4% relative to the original DiCoW on the EMMA MT-ASR benchmark.

Yiwen Shao, Yong Xu, Sanjeev Khudanpur et al.

Spatial information is a critical clue for multi-channel multi-speaker target speech recognition. Most state-of-the-art multi-channel Automatic Speech Recognition (ASR) systems extract spatial features only during the speech separation stage, followed by standard single-channel ASR on the separated speech. This approach results in an inefficient, lengthy pipeline and sub-optimal ASR performance due to the accumulated errors from preprocessing modules. Furthermore, most spatial feature extraction methods depend on the knowledge of speaker positions and microphone topology, making the systems reliant on specific settings and challenging to adapt to new equipment. In this work, we propose a solution to these issues with a lightweight embedding module named SpatialEmb, which extracts and encodes spatial information directly for the ASR model, supporting both fixed and arbitrary microphone topology. We conduct comprehensive experiments on AliMeeting, a real meeting corpus, to determine the optimal model design for SpatialEmb in terms of both performance and efficiency. Our best model trained with 105 hours Train-Ali-far achieves 17.04% and 20.32% character error rates (CER) on the Eval and Test sets, establishing a new state-of-the-art result with the same training data.

Dominik Klement, Matthew Maciejewski, Sanjeev Khudanpur et al.

The majority of deep learning-based speech enhancement methods require paired clean-noisy speech data. Collecting such data at scale in real-world conditions is infeasible, which has led the community to rely on synthetically generated noisy speech. However, this introduces a gap between the training and testing phases. In this work, we propose a novel dual-branch encoder-decoder architecture for unsupervised speech enhancement that separates the input into clean speech and residual noise. Adversarial training is employed to impose priors on each branch, defined by unpaired datasets of clean speech and, optionally, noise. Experimental results show that our method achieves performance comparable to leading unsupervised speech enhancement approaches. Furthermore, we demonstrate the critical impact of clean speech data selection on enhancement performance. In particular, our findings reveal that performance may appear overly optimistic when in-domain clean speech data are used for prior definition -- a practice adopted in previous unsupervised speech enhancement studies.

Cihan Xiao, Matthew Wiesner, Debashish Chakraborty et al.

Encoder-decoder models have achieved remarkable success in speech and text tasks, yet efficiently adapting these models to diverse uni/multi-modal scenarios remains an open challenge. In this paper, we propose Whisper-UT, a unified and efficient framework that leverages lightweight adapters to enable seamless adaptation across tasks, including a multi-modal machine translation (MMT) task that explicitly conditions translation on both speech and source language text inputs. By incorporating ASR hypotheses or ground-truth transcripts as prompts, this approach not only enables the system to process both modalities simultaneously but also enhances speech translation (ST) performance through a 2-stage decoding strategy. We demonstrate our methods using the Whisper model, though in principle they are general and could be applied to similar multitask models. We highlight the effectiveness of cross-modal and cross-task fine-tuning, which improves performance without requiring 3-way parallel data. Our results underscore the flexibility, efficiency, and general applicability of the proposed framework for multi-modal translation.

Amir Hussein, Cihan Xiao, Matthew Wiesner et al.

Neural transducers (NT) provide an effective framework for speech streaming, demonstrating strong performance in automatic speech recognition (ASR). However, the application of NT to speech translation (ST) remains challenging, as existing approaches struggle with word reordering and performance degradation when jointly modeling ASR and ST, resulting in a gap with attention-based encoder-decoder (AED) models. Existing NT-based ST approaches also suffer from high computational training costs. To address these issues, we propose HENT-SRT (Hierarchical Efficient Neural Transducer for Speech Recognition and Translation), a novel framework that factorizes ASR and translation tasks to better handle reordering. To ensure robust ST while preserving ASR performance, we use self-distillation with CTC consistency regularization. Moreover, we improve computational efficiency by incorporating best practices from ASR transducers, including a down-sampled hierarchical encoder, a stateless predictor, and a pruned transducer loss to reduce training complexity. Finally, we introduce a blank penalty during decoding, reducing deletions and improving translation quality. Our approach is evaluated on three conversational datasets Arabic, Spanish, and Mandarin achieving new state-of-the-art performance among NT models and substantially narrowing the gap with AED-based systems.

Cihan Xiao, Ruixing Liang, Xiangyu Zhang et al.

The success of large language models has driven interest in developing similar speech processing capabilities. However, a key challenge is the scarcity of high-quality spontaneous speech data, as most existing datasets contain scripted dialogues. To address this, we present a novel pipeline for eliciting and recording natural dialogues and release our dataset with 100+ hours of spontaneous speech. Our approach fosters fluid, natural conversations while encouraging a diverse range of topics and interactive exchanges. Unlike traditional methods, it facilitates genuine interactions, providing a reproducible framework for future data collection. This paper introduces our dataset and methodology, laying the groundwork for addressing the shortage of spontaneous speech data. We plan to expand this dataset in future stages, offering a growing resource for the research community.

Suzy J. Styles, Victoria Y. H. Chua, Fei Ting Woon et al.

Language development experts need tools that can automatically identify languages from fluent, conversational speech, and provide reliable estimates of usage rates at the level of an individual recording. However, language identification systems are typically evaluated on metrics such as equal error rate and balanced accuracy, applied at the level of an entire speech corpus. These overview metrics do not provide information about model performance at the level of individual speakers, recordings, or units of speech with different linguistic characteristics. Overview statistics may therefore mask systematic errors in model performance for some subsets of the data, and consequently, have worse performance on data derived from some subsets of human speakers, creating a kind of algorithmic bias. In the current paper, we investigate how well a number of language identification systems perform on individual recordings and speech units with different linguistic properties in the MERLIon CCS Challenge. The Challenge dataset features accented English-Mandarin code-switched child-directed speech.

Amir Hussein, Shammur Absar Chowdhury, Ahmed Abdelali et al.

The pervasiveness of intra-utterance code-switching (CS) in spoken content requires that speech recognition (ASR) systems handle mixed language. Designing a CS-ASR system has many challenges, mainly due to data scarcity, grammatical structure complexity, and domain mismatch. The most common method for addressing CS is to train an ASR system with the available transcribed CS speech, along with monolingual data. In this work, we propose a zero-shot learning methodology for CS-ASR by augmenting the monolingual data with artificially generating CS text. We based our approach on random lexical replacements and Equivalence Constraint (EC) while exploiting aligned translation pairs to generate random and grammatically valid CS content. Our empirical results show a 65.5% relative reduction in language model perplexity, and 7.7% in ASR WER on two ecologically valid CS test sets. The human evaluation of the generated text using EC suggests that more than 80% is of adequate quality.

Piotr Żelasko, Daniel Povey, Jan "Yenda" Trmal et al.

Speech data is notoriously difficult to work with due to a variety of codecs, lengths of recordings, and meta-data formats. We present Lhotse, a speech data representation library that draws upon lessons learned from Kaldi speech recognition toolkit and brings its concepts into the modern deep learning ecosystem. Lhotse provides a common JSON description format with corresponding Python classes and data preparation recipes for over 30 popular speech corpora. Various datasets can be easily combined together and re-purposed for different tasks. The library handles multi-channel recordings, long recordings, local and cloud storage, lazy and on-the-fly operations amongst other features. We introduce Cut and CutSet concepts, which simplify common data wrangling tasks for audio and help incorporate acoustic context of speech utterances. Finally, we show how Lhotse leverages PyTorch data API abstractions and adopts them to handle speech data for deep learning.

Matthew Wiesner, Desh Raj, Sanjeev Khudanpur

Self-supervised model pre-training has recently garnered significant interest, but relatively few efforts have explored using additional resources in fine-tuning these models. We demonstrate how universal phoneset acoustic models can leverage cross-lingual supervision to improve transfer of pretrained self-supervised representations to new languages. We also show how target-language text can be used to enable and improve fine-tuning with the lattice-free maximum mutual information (LF-MMI) objective. In three low-resource languages these techniques greatly improved few-shot learning performance.

Guoguo Chen, Shuzhou Chai, Guanbo Wang et al.

This paper introduces GigaSpeech, an evolving, multi-domain English speech recognition corpus with 10,000 hours of high quality labeled audio suitable for supervised training, and 40,000 hours of total audio suitable for semi-supervised and unsupervised training. Around 40,000 hours of transcribed audio is first collected from audiobooks, podcasts and YouTube, covering both read and spontaneous speaking styles, and a variety of topics, such as arts, science, sports, etc. A new forced alignment and segmentation pipeline is proposed to create sentence segments suitable for speech recognition training, and to filter out segments with low-quality transcription. For system training, GigaSpeech provides five subsets of different sizes, 10h, 250h, 1000h, 2500h, and 10000h. For our 10,000-hour XL training subset, we cap the word error rate at 4% during the filtering/validation stage, and for all our other smaller training subsets, we cap it at 0%. The DEV and TEST evaluation sets, on the other hand, are re-processed by professional human transcribers to ensure high transcription quality. Baseline systems are provided for popular speech recognition toolkits, namely Athena, ESPnet, Kaldi and Pika.

Piotr Żelasko, Sonal Joshi, Yiwen Shao et al.

The ubiquitous presence of machine learning systems in our lives necessitates research into their vulnerabilities and appropriate countermeasures. In particular, we investigate the effectiveness of adversarial attacks and defenses against automatic speech recognition (ASR) systems. We select two ASR models - a thoroughly studied DeepSpeech model and a more recent Espresso framework Transformer encoder-decoder model. We investigate two threat models: a denial-of-service scenario where fast gradient-sign method (FGSM) or weak projected gradient descent (PGD) attacks are used to degrade the model's word error rate (WER); and a targeted scenario where a more potent imperceptible attack forces the system to recognize a specific phrase. We find that the attack transferability across the investigated ASR systems is limited. To defend the model, we use two preprocessing defenses: randomized smoothing and WaveGAN-based vocoder, and find that they significantly improve the model's adversarial robustness. We show that a WaveGAN vocoder can be a useful countermeasure to adversarial attacks on ASR systems - even when it is jointly attacked with the ASR, the target phrases' word error rate is high.

Hang Lv, Zhehuai Chen, Hainan Xu et al.

We introduce asynchronous dynamic decoder, which adopts an efficient A* algorithm to incorporate big language models in the one-pass decoding for large vocabulary continuous speech recognition. Unlike standard one-pass decoding with on-the-fly composition decoder which might induce a significant computation overhead, the asynchronous dynamic decoder has a novel design where it has two fronts, with one performing "exploration" and the other "backfill". The computation of the two fronts alternates in the decoding process, resulting in more effective pruning than the standard one-pass decoding with an on-the-fly composition decoder. Experiments show that the proposed decoder works notably faster than the standard one-pass decoding with on-the-fly composition decoder, while the acceleration will be more obvious with the increment of data complexity.

Gaurav Kumar, Philipp Koehn, Sanjeev Khudanpur

Large web-crawled corpora represent an excellent resource for improving the performance of Neural Machine Translation (NMT) systems across several language pairs. However, since these corpora are typically extremely noisy, their use is fairly limited. Current approaches to dealing with this problem mainly focus on filtering using heuristics or single features such as language model scores or bi-lingual similarity. This work presents an alternative approach which learns weights for multiple sentence-level features. These feature weights which are optimized directly for the task of improving translation performance, are used to score and filter sentences in the noisy corpora more effectively. We provide results of applying this technique to building NMT systems using the Paracrawl corpus for Estonian-English and show that it beats strong single feature baselines and hand designed combinations. Additionally, we analyze the sensitivity of this method to different types of noise and explore if the learned weights generalize to other language pairs using the Maltese-English Paracrawl corpus.

Gaurav Kumar, Philipp Koehn, Sanjeev Khudanpur

Low-resource Multilingual Neural Machine Translation (MNMT) is typically tasked with improving the translation performance on one or more language pairs with the aid of high-resource language pairs. In this paper, we propose two simple search based curricula -- orderings of the multilingual training data -- which help improve translation performance in conjunction with existing techniques such as fine-tuning. Additionally, we attempt to learn a curriculum for MNMT from scratch jointly with the training of the translation system with the aid of contextual multi-arm bandits. We show on the FLORES low-resource translation dataset that these learned curricula can provide better starting points for fine tuning and improve overall performance of the translation system.

Ke Li, Daniel Povey, Sanjeev Khudanpur

This paper proposes a parallel computation strategy and a posterior-based lattice expansion algorithm for efficient lattice rescoring with neural language models (LMs) for automatic speech recognition. First, lattices from first-pass decoding are expanded by the proposed posterior-based lattice expansion algorithm. Second, each expanded lattice is converted into a minimal list of hypotheses that covers every arc. Each hypothesis is constrained to be the best path for at least one arc it includes. For each lattice, the neural LM scores of the minimal list are computed in parallel and are then integrated back to the lattice in the rescoring stage. Experiments on the Switchboard dataset show that the proposed rescoring strategy obtains comparable recognition performance and generates more compact lattices than a competitive baseline method. Furthermore, the parallel rescoring method offers more flexibility by simplifying the integration of PyTorch-trained neural LMs for lattice rescoring with Kaldi.

Yiming Wang, Hang Lv, Daniel Povey et al.

Modern wake word detection systems usually rely on neural networks for acoustic modeling. Transformers has recently shown superior performance over LSTM and convolutional networks in various sequence modeling tasks with their better temporal modeling power. However it is not clear whether this advantage still holds for short-range temporal modeling like wake word detection. Besides, the vanilla Transformer is not directly applicable to the task due to its non-streaming nature and the quadratic time and space complexity. In this paper we explore the performance of several variants of chunk-wise streaming Transformers tailored for wake word detection in a recently proposed LF-MMI system, including looking-ahead to the next chunk, gradient stopping, different positional embedding methods and adding same-layer dependency between chunks. Our experiments on the Mobvoi wake word dataset demonstrate that our proposed Transformer model outperforms the baseline convolution network by 25% on average in false rejection rate at the same false alarm rate with a comparable model size, while still maintaining linear complexity w.r.t. the sequence length.

Shota Horiguchi, Nelson Yalta, Paola Garcia et al.

This paper provides a detailed description of the Hitachi-JHU system that was submitted to the Third DIHARD Speech Diarization Challenge. The system outputs the ensemble results of the five subsystems: two x-vector-based subsystems, two end-to-end neural diarization-based subsystems, and one hybrid subsystem. We refine each system and all five subsystems become competitive and complementary. After the DOVER-Lap based system combination, it achieved diarization error rates of 11.58 % and 14.09 % in Track 1 full and core, and 16.94 % and 20.01 % in Track 2 full and core, respectively. With their results, we won second place in all the tasks of the challenge.

Jonathan D. Jones, Cathryn Cortesa, Amy Shelton et al.

In this paper we address the task of recognizing assembly actions as a structure (e.g. a piece of furniture or a toy block tower) is built up from a set of primitive objects. Recognizing the full range of assembly actions requires perception at a level of spatial detail that has not been attempted in the action recognition literature to date. We extend the fine-grained activity recognition setting to address the task of assembly action recognition in its full generality by unifying assembly actions and kinematic structures within a single framework. We use this framework to develop a general method for recognizing assembly actions from observation sequences, along with observation features that take advantage of a spatial assembly's special structure. Finally, we evaluate our method empirically on two application-driven data sources: (1) An IKEA furniture-assembly dataset, and (2) A block-building dataset. On the first, our system recognizes assembly actions with an average framewise accuracy of 70% and an average normalized edit distance of 10%. On the second, which requires fine-grained geometric reasoning to distinguish between assemblies, our system attains an average normalized edit distance of 23% -- a relative improvement of 69% over prior work.

Desh Raj, Zili Huang, Sanjeev Khudanpur

This paper describes a method for overlap-aware speaker diarization. Given an overlap detector and a speaker embedding extractor, our method performs spectral clustering of segments informed by the output of the overlap detector. This is achieved by transforming the discrete clustering problem into a convex optimization problem which is solved by eigen-decomposition. Thereafter, we discretize the solution by alternatively using singular value decomposition and a modified version of non-maximal suppression which is constrained by the output of the overlap detector. Furthermore, we detail an HMM-DNN based overlap detector which performs frame-level classification and enforces duration constraints through HMM state transitions. Our method achieves a test diarization error rate (DER) of 24.0% on the mixed-headset setting of the AMI meeting corpus, which is a relative improvement of 15.2% over a strong agglomerative hierarchical clustering baseline, and compares favorably with other overlap-aware diarization methods. Further analysis on the LibriCSS data demonstrates the effectiveness of the proposed method in high overlap conditions.

Desh Raj, Jesus Villalba, Daniel Povey et al.

Environmental noises and reverberation have a detrimental effect on the performance of automatic speech recognition (ASR) systems. Multi-condition training of neural network-based acoustic models is used to deal with this problem, but it requires many-folds data augmentation, resulting in increased training time. In this paper, we propose utterance-level noise vectors for noise-aware training of acoustic models in hybrid ASR. Our noise vectors are obtained by combining the means of speech frames and silence frames in the utterance, where the speech/silence labels may be obtained from a GMM-HMM model trained for ASR alignments, such that no extra computation is required beyond averaging of feature vectors. We show through experiments on AMI and Aurora-4 that this simple adaptation technique can result in 6-7% relative WER improvement. We implement several embedding-based adaptation baselines proposed in literature, and show that our method outperforms them on both the datasets. Finally, we extend our method to the online ASR setting by using frame-level maximum likelihood for the mean estimation.

Desh Raj, Leibny Paola Garcia-Perera, Zili Huang et al.

Several advances have been made recently towards handling overlapping speech for speaker diarization. Since speech and natural language tasks often benefit from ensemble techniques, we propose an algorithm for combining outputs from such diarization systems through majority voting. Our method, DOVER-Lap, is inspired from the recently proposed DOVER algorithm, but is designed to handle overlapping segments in diarization outputs. We also modify the pair-wise incremental label mapping strategy used in DOVER, and propose an approximation algorithm based on weighted k-partite graph matching, which performs this mapping using a global cost tensor. We demonstrate the strength of our method by combining outputs from diverse systems -- clustering-based, region proposal networks, and target-speaker voice activity detection -- on AMI and LibriCSS datasets, where it consistently outperforms the single best system. Additionally, we show that DOVER-Lap can be used for late fusion in multichannel diarization, and compares favorably with early fusion methods like beamforming.

Matthew Maciejewski, Jing Shi, Shinji Watanabe et al.

As the performance of single-channel speech separation systems has improved, there has been a desire to move to more challenging conditions than the clean, near-field speech that initial systems were developed on. When training deep learning separation models, a need for ground truth leads to training on synthetic mixtures. As such, training in noisy conditions requires either using noise synthetically added to clean speech, preventing the use of in-domain data for a noisy-condition task, or training using mixtures of noisy speech, requiring the network to additionally separate the noise. We demonstrate the relative inseparability of noise and that this noisy speech paradigm leads to significant degradation of system performance. We also propose an SI-SDR-inspired training objective that tries to exploit the inseparability of noise to implicitly partition the signal and discount noise separation errors, enabling the training of better separation systems with noisy oracle sources.

Ruizhe Huang, Ke Li, Ashish Arora et al.

This paper presents an efficient algorithm for n-gram language model adaptation under the minimum discrimination information (MDI) principle, where an out-of-domain language model is adapted to satisfy the constraints of marginal probabilities of the in-domain data. The challenge for MDI language model adaptation is its computational complexity. By taking advantage of the backoff structure of n-gram model and the idea of hierarchical training method, originally proposed for maximum entropy (ME) language models, we show that MDI adaptation can be computed in linear-time complexity to the inputs in each iteration. The complexity remains the same as ME models, although MDI is more general than ME. This makes MDI adaptation practical for large corpus and vocabulary. Experimental results confirm the scalability of our algorithm on very large datasets, while MDI adaptation gets slightly worse perplexity but better word error rate results compared to simple linear interpolation.