-
F-Actor: Controllable Conversational Behaviour in Full-Duplex Models
Authors:
Maike Züfle,
Ondrej Klejch,
Nicholas Sanders,
Jan Niehues,
Alexandra Birch,
Tsz Kin Lam
Abstract:
Spoken conversational systems require more than accurate speech generation to have human-like conversations: to feel natural and engaging, they must produce conversational behaviour that adapts dynamically to the context. Current spoken conversational systems, however, rarely allow such customization, limiting their naturalness and usability. In this work, we present the first open, instruction-fo…
▽ More
Spoken conversational systems require more than accurate speech generation to have human-like conversations: to feel natural and engaging, they must produce conversational behaviour that adapts dynamically to the context. Current spoken conversational systems, however, rarely allow such customization, limiting their naturalness and usability. In this work, we present the first open, instruction-following full-duplex conversational speech model that can be trained efficiently under typical academic resource constraints. By keeping the audio encoder frozen and finetuning only the language model, our model requires just 2,000 hours of data, without relying on large-scale pretraining or multi-stage optimization. The model can follow explicit instructions to control speaker voice, conversation topic, conversational behaviour (e.g., backchanneling and interruptions), and dialogue initiation. We propose a single-stage training protocol and systematically analyze design choices. Both the model and training code is released to enable reproducible research on controllable full-duplex speech systems.
△ Less
Submitted 15 April, 2026; v1 submitted 16 January, 2026;
originally announced January 2026.
-
The Prosody of Emojis
Authors:
Giulio Zhou,
Tsz Kin Lam,
Alexandra Birch,
Barry Haddow
Abstract:
Prosodic features such as pitch, timing, and intonation are central to spoken communication, conveying emotion, intent, and discourse structure. In text-based settings, where these cues are absent, emojis act as visual surrogates that add affective and pragmatic nuance. This study examines how emojis influence prosodic realisation in speech and how listeners interpret prosodic cues to recover emoj…
▽ More
Prosodic features such as pitch, timing, and intonation are central to spoken communication, conveying emotion, intent, and discourse structure. In text-based settings, where these cues are absent, emojis act as visual surrogates that add affective and pragmatic nuance. This study examines how emojis influence prosodic realisation in speech and how listeners interpret prosodic cues to recover emoji meanings. Unlike previous work, we directly link prosody and emojis by analysing human speech data collected through a controlled elicited production task. Using Bayesian multilevel modelling, we show that speakers systematically adapt their prosody based on emoji cues, and that listeners can recover intended meanings significantly above chance. Furthermore, our results reveal a clear hierarchy in prosodic shifts: greater semantic differences between emojis correspond to increased prosodic divergence. These findings suggest that emojis are meaningful carriers of prosodic intent that bridge the gap between digital text and spoken production.
△ Less
Submitted 4 June, 2026; v1 submitted 1 August, 2025;
originally announced August 2025.
-
From TOWER to SPIRE: Adding the Speech Modality to a Translation-Specialist LLM
Authors:
Kshitij Ambilduke,
Ben Peters,
Sonal Sannigrahi,
Anil Keshwani,
Tsz Kin Lam,
Bruno Martins,
André F. T. Martins,
Marcely Zanon Boito
Abstract:
We introduce Spire, a speech-augmented language model (LM) capable of both translating and transcribing speech input from English into 10 other languages as well as translating text input in both language directions. Spire integrates the speech modality into an existing multilingual LM via speech discretization and continued pre-training using only 42.5K hours of speech. In particular, we adopt th…
▽ More
We introduce Spire, a speech-augmented language model (LM) capable of both translating and transcribing speech input from English into 10 other languages as well as translating text input in both language directions. Spire integrates the speech modality into an existing multilingual LM via speech discretization and continued pre-training using only 42.5K hours of speech. In particular, we adopt the pretraining framework of multilingual LMs and treat discretized speech input as an additional translation language. This approach not only equips the model with speech capabilities, but also preserves its strong text-based performance. We achieve this using significantly less data than existing speech LMs, demonstrating that discretized speech input integration as an additional language is feasible during LM adaptation. We make our code and models available to the community.
△ Less
Submitted 22 October, 2025; v1 submitted 13 March, 2025;
originally announced March 2025.
-
Prepending or Cross-Attention for Speech-to-Text? An Empirical Comparison
Authors:
Tsz Kin Lam,
Marco Gaido,
Sara Papi,
Luisa Bentivogli,
Barry Haddow
Abstract:
Following the remarkable success of Large Language Models (LLMs) in NLP tasks, there is increasing interest in extending their capabilities to speech -- the most common form of communication. The most widespread approach to integrating speech into LLMs is dense feature prepending (DFP), which prepends the projected speech representations to the textual representations, allowing end-to-end training…
▽ More
Following the remarkable success of Large Language Models (LLMs) in NLP tasks, there is increasing interest in extending their capabilities to speech -- the most common form of communication. The most widespread approach to integrating speech into LLMs is dense feature prepending (DFP), which prepends the projected speech representations to the textual representations, allowing end-to-end training with a speech encoder. This raises questions about the need for a sophisticated speech encoder for DFP and how its performance compares with a standard encoder-decoder (i.e., cross-attention) architecture. We compare DFP and cross-attention under a variety of configurations, such as CTC compression, sequence-level knowledge distillation, on monolingual, bilingual, and multilingual models. To perform a controlled architectural comparison, we train all models from scratch rather than using large pretrained models and use comparable data and parameter settings, testing speech-to-text recognition (ASR) and translation (ST) on MuST-C v1.0 and CoVoST2 datasets. Despite the wide adoption of DFP, our results do not indicate a clear advantage of DFP over cross-attention.
△ Less
Submitted 7 February, 2025; v1 submitted 4 January, 2025;
originally announced January 2025.
-
Findings of the IWSLT 2024 Evaluation Campaign
Authors:
Ibrahim Said Ahmad,
Antonios Anastasopoulos,
Ondřej Bojar,
Claudia Borg,
Marine Carpuat,
Roldano Cattoni,
Mauro Cettolo,
William Chen,
Qianqian Dong,
Marcello Federico,
Barry Haddow,
Dávid Javorský,
Mateusz Krubiński,
Tsz Kin Lam,
Xutai Ma,
Prashant Mathur,
Evgeny Matusov,
Chandresh Maurya,
John McCrae,
Kenton Murray,
Satoshi Nakamura,
Matteo Negri,
Jan Niehues,
Xing Niu,
Atul Kr. Ojha
, et al. (20 additional authors not shown)
Abstract:
This paper reports on the shared tasks organized by the 21st IWSLT Conference. The shared tasks address 7 scientific challenges in spoken language translation: simultaneous and offline translation, automatic subtitling and dubbing, speech-to-speech translation, dialect and low-resource speech translation, and Indic languages. The shared tasks attracted 18 teams whose submissions are documented in…
▽ More
This paper reports on the shared tasks organized by the 21st IWSLT Conference. The shared tasks address 7 scientific challenges in spoken language translation: simultaneous and offline translation, automatic subtitling and dubbing, speech-to-speech translation, dialect and low-resource speech translation, and Indic languages. The shared tasks attracted 18 teams whose submissions are documented in 26 system papers. The growing interest towards spoken language translation is also witnessed by the constantly increasing number of shared task organizers and contributors to the overview paper, almost evenly distributed across industry and academia.
△ Less
Submitted 7 November, 2024;
originally announced November 2024.
-
Pitfalls and Outlooks in Using COMET
Authors:
Vilém Zouhar,
Pinzhen Chen,
Tsz Kin Lam,
Nikita Moghe,
Barry Haddow
Abstract:
The COMET metric has blazed a trail in the machine translation community, given its strong correlation with human judgements of translation quality. Its success stems from being a modified pre-trained multilingual model finetuned for quality assessment. However, it being a machine learning model also gives rise to a new set of pitfalls that may not be widely known. We investigate these unexpected…
▽ More
The COMET metric has blazed a trail in the machine translation community, given its strong correlation with human judgements of translation quality. Its success stems from being a modified pre-trained multilingual model finetuned for quality assessment. However, it being a machine learning model also gives rise to a new set of pitfalls that may not be widely known. We investigate these unexpected behaviours from three aspects: 1) technical: obsolete software versions and compute precision; 2) data: empty content, language mismatch, and translationese at test time as well as distribution and domain biases in training; 3) usage and reporting: multi-reference support and model referencing in the literature. All of these problems imply that COMET scores are not comparable between papers or even technical setups and we put forward our perspective on fixing each issue. Furthermore, we release the sacreCOMET package that can generate a signature for the software and model configuration as well as an appropriate citation. The goal of this work is to help the community make more sound use of the COMET metric.
△ Less
Submitted 30 September, 2024; v1 submitted 27 August, 2024;
originally announced August 2024.
-
Compact Speech Translation Models via Discrete Speech Units Pretraining
Authors:
Tsz Kin Lam,
Alexandra Birch,
Barry Haddow
Abstract:
We propose a pretraining method to use Self-Supervised Speech (SSS) model to creating more compact Speech-to-text Translation. In contrast to using the SSS model for initialization, our method is more suitable to memory constrained scenario such as on-device deployment. Our method is based on Discrete Speech Units (DSU) extracted from the SSS model. In the first step, our method pretrains two smal…
▽ More
We propose a pretraining method to use Self-Supervised Speech (SSS) model to creating more compact Speech-to-text Translation. In contrast to using the SSS model for initialization, our method is more suitable to memory constrained scenario such as on-device deployment. Our method is based on Discrete Speech Units (DSU) extracted from the SSS model. In the first step, our method pretrains two smaller encoder-decoder models on 1) Filterbank-to-DSU (Fbk-to-DSU) and 2) DSU-to-Translation (DSU-to-Trl) data respectively. The DSU thus become the distillation inputs of the smaller models. Subsequently, the encoder from the Fbk-to-DSU model and the decoder from the DSU-to-Trl model are taken to initialise the compact model. Finally, the compact model is finetuned on the paired Fbk-Trl data. In addition to being compact, our method requires no transcripts, making it applicable to low-resource settings. It also avoids speech discretization in inference and is more robust to the DSU tokenization. Evaluation on CoVoST-2 (X-En) shows that our method has consistent improvement over the baseline in three metrics while being compact i.e., only half the SSS model size.
△ Less
Submitted 26 June, 2024; v1 submitted 29 February, 2024;
originally announced February 2024.
-
Prosody in Cascade and Direct Speech-to-Text Translation: a case study on Korean Wh-Phrases
Authors:
Giulio Zhou,
Tsz Kin Lam,
Alexandra Birch,
Barry Haddow
Abstract:
Speech-to-Text Translation (S2TT) has typically been addressed with cascade systems, where speech recognition systems generate a transcription that is subsequently passed to a translation model. While there has been a growing interest in developing direct speech translation systems to avoid propagating errors and losing non-verbal content, prior work in direct S2TT has struggled to conclusively es…
▽ More
Speech-to-Text Translation (S2TT) has typically been addressed with cascade systems, where speech recognition systems generate a transcription that is subsequently passed to a translation model. While there has been a growing interest in developing direct speech translation systems to avoid propagating errors and losing non-verbal content, prior work in direct S2TT has struggled to conclusively establish the advantages of integrating the acoustic signal directly into the translation process. This work proposes using contrastive evaluation to quantitatively measure the ability of direct S2TT systems to disambiguate utterances where prosody plays a crucial role. Specifically, we evaluated Korean-English translation systems on a test set containing wh-phrases, for which prosodic features are necessary to produce translations with the correct intent, whether it's a statement, a yes/no question, a wh-question, and more. Our results clearly demonstrate the value of direct translation systems over cascade translation models, with a notable 12.9% improvement in overall accuracy in ambiguous cases, along with up to a 15.6% increase in F1 scores for one of the major intent categories. To the best of our knowledge, this work stands as the first to provide quantitative evidence that direct S2TT models can effectively leverage prosody. The code for our evaluation is openly accessible and freely available for review and utilisation.
△ Less
Submitted 1 February, 2024;
originally announced February 2024.
-
Futuristic Variations and Analysis in Fundus Images Corresponding to Biological Traits
Authors:
Muhammad Hassan,
Hao Zhang,
Ahmed Fateh Ameen,
Home Wu Zeng,
Shuye Ma,
Wen Liang,
Dingqi Shang,
Jiaming Ding,
Ziheng Zhan,
Tsz Kwan Lam,
Ming Xu,
Qiming Huang,
Dongmei Wu,
Can Yang Zhang,
Zhou You,
Awiwu Ain,
Pei Wu Qin
Abstract:
Fundus image captures rear of an eye, and which has been studied for the diseases identification, classification, segmentation, generation, and biological traits association using handcrafted, conventional, and deep learning methods. In biological traits estimation, most of the studies have been carried out for the age prediction and gender classification with convincing results. However, the curr…
▽ More
Fundus image captures rear of an eye, and which has been studied for the diseases identification, classification, segmentation, generation, and biological traits association using handcrafted, conventional, and deep learning methods. In biological traits estimation, most of the studies have been carried out for the age prediction and gender classification with convincing results. However, the current study utilizes the cutting-edge deep learning (DL) algorithms to estimate biological traits in terms of age and gender together with associating traits to retinal visuals. For the traits association, our study embeds aging as the label information into the proposed DL model to learn knowledge about the effected regions with aging. Our proposed DL models, named FAG-Net and FGC-Net, correspondingly estimate biological traits (age and gender) and generates fundus images. FAG-Net can generate multiple variants of an input fundus image given a list of ages as conditions. Our study analyzes fundus images and their corresponding association with biological traits, and predicts of possible spreading of ocular disease on fundus images given age as condition to the generative model. Our proposed models outperform the randomly selected state of-the-art DL models.
△ Less
Submitted 7 February, 2023;
originally announced February 2023.
-
Make More of Your Data: Minimal Effort Data Augmentation for Automatic Speech Recognition and Translation
Authors:
Tsz Kin Lam,
Shigehiko Schamoni,
Stefan Riezler
Abstract:
Data augmentation is a technique to generate new training data based on existing data. We evaluate the simple and cost-effective method of concatenating the original data examples to build new training instances. Continued training with such augmented data is able to improve off-the-shelf Transformer and Conformer models that were optimized on the original data only. We demonstrate considerable im…
▽ More
Data augmentation is a technique to generate new training data based on existing data. We evaluate the simple and cost-effective method of concatenating the original data examples to build new training instances. Continued training with such augmented data is able to improve off-the-shelf Transformer and Conformer models that were optimized on the original data only. We demonstrate considerable improvements on the LibriSpeech-960h test sets (WER 2.83 and 6.87 for test-clean and test-other), which carry over to models combined with shallow fusion (WER 2.55 and 6.27). Our method of continued training also leads to improvements of up to 0.9 WER on the ASR part of CoVoST-2 for four non English languages, and we observe that the gains are highly dependent on the size of the original training data. We compare different concatenation strategies and found that our method does not need speaker information to achieve its improvements. Finally, we demonstrate on two datasets that our methods also works for speech translation tasks.
△ Less
Submitted 14 April, 2023; v1 submitted 27 October, 2022;
originally announced October 2022.
-
Analyzing the Use of Influence Functions for Instance-Specific Data Filtering in Neural Machine Translation
Authors:
Tsz Kin Lam,
Eva Hasler,
Felix Hieber
Abstract:
Customer feedback can be an important signal for improving commercial machine translation systems. One solution for fixing specific translation errors is to remove the related erroneous training instances followed by re-training of the machine translation system, which we refer to as instance-specific data filtering. Influence functions (IF) have been shown to be effective in finding such relevant…
▽ More
Customer feedback can be an important signal for improving commercial machine translation systems. One solution for fixing specific translation errors is to remove the related erroneous training instances followed by re-training of the machine translation system, which we refer to as instance-specific data filtering. Influence functions (IF) have been shown to be effective in finding such relevant training examples for classification tasks such as image classification, toxic speech detection and entailment task. Given a probing instance, IF find influential training examples by measuring the similarity of the probing instance with a set of training examples in gradient space. In this work, we examine the use of influence functions for Neural Machine Translation (NMT). We propose two effective extensions to a state of the art influence function and demonstrate on the sub-problem of copied training examples that IF can be applied more generally than handcrafted regular expressions.
△ Less
Submitted 24 October, 2022;
originally announced October 2022.
-
Sample, Translate, Recombine: Leveraging Audio Alignments for Data Augmentation in End-to-end Speech Translation
Authors:
Tsz Kin Lam,
Shigehiko Schamoni,
Stefan Riezler
Abstract:
End-to-end speech translation relies on data that pair source-language speech inputs with corresponding translations into a target language. Such data are notoriously scarce, making synthetic data augmentation by back-translation or knowledge distillation a necessary ingredient of end-to-end training. In this paper, we present a novel approach to data augmentation that leverages audio alignments,…
▽ More
End-to-end speech translation relies on data that pair source-language speech inputs with corresponding translations into a target language. Such data are notoriously scarce, making synthetic data augmentation by back-translation or knowledge distillation a necessary ingredient of end-to-end training. In this paper, we present a novel approach to data augmentation that leverages audio alignments, linguistic properties, and translation. First, we augment a transcription by sampling from a suffix memory that stores text and audio data. Second, we translate the augmented transcript. Finally, we recombine concatenated audio segments and the generated translation. Besides training an MT-system, we only use basic off-the-shelf components without fine-tuning. While having similar resource demands as knowledge distillation, adding our method delivers consistent improvements of up to 0.9 and 1.1 BLEU points on five language pairs on CoVoST 2 and on two language pairs on Europarl-ST, respectively.
△ Less
Submitted 16 March, 2022;
originally announced March 2022.
-
On-the-Fly Aligned Data Augmentation for Sequence-to-Sequence ASR
Authors:
Tsz Kin Lam,
Mayumi Ohta,
Shigehiko Schamoni,
Stefan Riezler
Abstract:
We propose an on-the-fly data augmentation method for automatic speech recognition (ASR) that uses alignment information to generate effective training samples. Our method, called Aligned Data Augmentation (ADA) for ASR, replaces transcribed tokens and the speech representations in an aligned manner to generate previously unseen training pairs. The speech representations are sampled from an audio…
▽ More
We propose an on-the-fly data augmentation method for automatic speech recognition (ASR) that uses alignment information to generate effective training samples. Our method, called Aligned Data Augmentation (ADA) for ASR, replaces transcribed tokens and the speech representations in an aligned manner to generate previously unseen training pairs. The speech representations are sampled from an audio dictionary that has been extracted from the training corpus and inject speaker variations into the training examples. The transcribed tokens are either predicted by a language model such that the augmented data pairs are semantically close to the original data, or randomly sampled. Both strategies result in training pairs that improve robustness in ASR training. Our experiments on a Seq-to-Seq architecture show that ADA can be applied on top of SpecAugment, and achieves about 9-23% and 4-15% relative improvements in WER over SpecAugment alone on LibriSpeech 100h and LibriSpeech 960h test datasets, respectively.
△ Less
Submitted 9 June, 2021; v1 submitted 3 April, 2021;
originally announced April 2021.
-
Cascaded Models With Cyclic Feedback For Direct Speech Translation
Authors:
Tsz Kin Lam,
Shigehiko Schamoni,
Stefan Riezler
Abstract:
Direct speech translation describes a scenario where only speech inputs and corresponding translations are available. Such data are notoriously limited. We present a technique that allows cascades of automatic speech recognition (ASR) and machine translation (MT) to exploit in-domain direct speech translation data in addition to out-of-domain MT and ASR data. After pre-training MT and ASR, we use…
▽ More
Direct speech translation describes a scenario where only speech inputs and corresponding translations are available. Such data are notoriously limited. We present a technique that allows cascades of automatic speech recognition (ASR) and machine translation (MT) to exploit in-domain direct speech translation data in addition to out-of-domain MT and ASR data. After pre-training MT and ASR, we use a feedback cycle where the downstream performance of the MT system is used as a signal to improve the ASR system by self-training, and the MT component is fine-tuned on multiple ASR outputs, making it more tolerant towards spelling variations. A comparison to end-to-end speech translation using components of identical architecture and the same data shows gains of up to 3.8 BLEU points on LibriVoxDeEn and up to 5.1 BLEU points on CoVoST for German-to-English speech translation.
△ Less
Submitted 11 February, 2021; v1 submitted 21 October, 2020;
originally announced October 2020.
-
Interactive-Predictive Neural Machine Translation through Reinforcement and Imitation
Authors:
Tsz Kin Lam,
Shigehiko Schamoni,
Stefan Riezler
Abstract:
We propose an interactive-predictive neural machine translation framework for easier model personalization using reinforcement and imitation learning. During the interactive translation process, the user is asked for feedback on uncertain locations identified by the system. Responses are weak feedback in the form of "keep" and "delete" edits, and expert demonstrations in the form of "substitute" e…
▽ More
We propose an interactive-predictive neural machine translation framework for easier model personalization using reinforcement and imitation learning. During the interactive translation process, the user is asked for feedback on uncertain locations identified by the system. Responses are weak feedback in the form of "keep" and "delete" edits, and expert demonstrations in the form of "substitute" edits. Conditioning on the collected feedback, the system creates alternative translations via constrained beam search. In simulation experiments on two language pairs our systems get close to the performance of supervised training with much less human effort.
△ Less
Submitted 5 July, 2019; v1 submitted 4 July, 2019;
originally announced July 2019.
-
A Reinforcement Learning Approach to Interactive-Predictive Neural Machine Translation
Authors:
Tsz Kin Lam,
Julia Kreutzer,
Stefan Riezler
Abstract:
We present an approach to interactive-predictive neural machine translation that attempts to reduce human effort from three directions: Firstly, instead of requiring humans to select, correct, or delete segments, we employ the idea of learning from human reinforcements in form of judgments on the quality of partial translations. Secondly, human effort is further reduced by using the entropy of wor…
▽ More
We present an approach to interactive-predictive neural machine translation that attempts to reduce human effort from three directions: Firstly, instead of requiring humans to select, correct, or delete segments, we employ the idea of learning from human reinforcements in form of judgments on the quality of partial translations. Secondly, human effort is further reduced by using the entropy of word predictions as uncertainty criterion to trigger feedback requests. Lastly, online updates of the model parameters after every interaction allow the model to adapt quickly. We show in simulation experiments that reward signals on partial translations significantly improve character F-score and BLEU compared to feedback on full translations only, while human effort can be reduced to an average number of $5$ feedback requests for every input.
△ Less
Submitted 5 June, 2018; v1 submitted 3 May, 2018;
originally announced May 2018.