arXiv subjects

Gustav Eje Henter

Publications and source records attributed to Gustav Eje Henter.

At least 19 recordsLinked to original sources

arXiv2026

Iterative Self-Learning for Expressive Text-to-Speech Synthesis

Expressive text-to-speech (TTS) systems that use explicit conditioning labels provide direct and interpretable control over expressive attributes, in contrast to reference-based or prompting-based approaches, but require labeled data. Obtaining these labels at scale is costly and time-consuming, yet no prior semi-supervised framework addresses this specific bottleneck. Existing semi-supervised TTS methods instead target scarcity of paired speech-text data or transcriptions. To address the scarcity of expressive labels, we propose an Iterative Self-Learning (ISL) framework for expressive TTS, built on Invert-Classify, a classifier-free method that recovers discrete expressive labels by inverting a frozen generative model. The framework iteratively pseudo-labels unlabeled speech using the current model, retrains on the combined labeled and pseudo-labeled data, and repeats, progressively refining label quality and synthesis. We validate on two expressive tasks, word-level prominence and utterance-level emotion, across multiple low-resource data splits. We find that iterative refinement can improve pseudo-label accuracy over single-pass baselines. Furthermore, we observe that these improvements in pseudo-labeling of expressivity translate to gains in expressive label adherence and synthesis quality, confirmed by objective metrics and human listening tests. In the most data-scarce conditions, ISL-trained models outperform single-pass pseudo-labeling and further approach fully supervised performance, demonstrating that gradient-based ISL is an effective solution to expressive label scarcity in low-resource TTS.

eess.AS↗

arXiv2026

The GENEA Challenge 2026: A Large-Scale Disentangled Evaluation of Speech-Driven Gesture Generation on the Seamless Interaction Dataset

This preprint presents the results of the fourth GENEA Challenge, a large-scale human evaluation of five speech-driven gesture-generation systems trained by participating teams on the Seamless Interaction dataset of dyadic conversations. As in the 2023 GENEA Challenge, we used a disentangled evaluation methodology to assess motion quality and speech alignment without confounding between the two, and performed a dyadic mismatching study to isolate the effect of listening and reacting to the interlocutor. We additionally introduce a new semantic gesture-generation task and a text-mismatching evaluation methodology using the Grounded Gestures subset of the data. In total, we ran four large-scale user studies, collecting over 23,000 votes from 869 test-takers. In the motion-realism study, the dataset's filtered segments had substantially higher motion quality than all challenge submissions (68-95% pairwise winrate). In the speech-alignment study, the motion-capture segments provided a conceptual ceiling at 62% alignment score, with the top submission significantly behind at 32% and the rest only slightly above the 0% expected of an input-independent system. In the dyadic study, motion capture again set the ceiling at 65% appropriateness score, but no submission scored substantially above chance, indicating that the systems could not yet respond to the interlocutor. Finally, the semantic mismatching evaluation found highly expressive gestures in the dataset (test-takers identified the matching transcript 79% of the time), yet almost all submissions failed to generate semantically expressive motion, with the best achieving only an 8% appropriateness score. The collected votes and outputs will be made publicly available at https://genea-workshop.github.io/2026/challenge/ to facilitate reproducibility and further research.

cs.CV↗

arXiv2026

Causality for Tabular Data Synthesis: A High-Order Structure Causal Benchmark Framework

Existing evaluations of tabular synthesis models rely primarily on low-order statistics and downstream task performance, leaving multivariate causal relationships that go beyond pairwise correlations largely unmeasured. We argue that a systematic evaluation on high-order structural information is a crucial first step in addressing this issue in tabular data synthesis. In this paper, we present high-order structural causal information as a natural form of prior knowledge and introduce a benchmark framework to evaluate tabular synthesis models. This framework allows us to generate benchmark datasets through a flexible range of data generation processes, allowing for the training of tabular synthesis models using these datasets for further evaluation. We propose multiple benchmark tasks, high-order metrics, and causal inference tasks as downstream tasks for evaluating the quality of synthetic data generated by the trained models. Our experiments demonstrate the effectiveness of the benchmark framework in evaluating the model's ability to capture high-order structural causal information. Furthermore, our benchmarking results provide an initial assessment of state-of-the-art tabular synthesis models. These results reveal significant gaps between ideal and actual performance and highlight how baseline methods differ. We position the framework as a controlled diagnostic benchmark for causal fidelity, complementing existing low-order and downstream evaluations. We open source the benchmark framework, including both code and data along with documentation, to support further research in this area.

cs.LG↗

arXiv2026

The Voice Behind the Words: Quantifying Intersectional Bias in SpeechLLMs

Speech Large Language Models (SpeechLLMs) process spoken input directly, retaining cues such as accent and perceived gender that were previously removed in cascaded pipelines. This introduces speaker identity dependent variation in responses. We present a large-scale intersectional evaluation of accent and gender bias in three SpeechLLMs using 2,880 controlled interactions across six English accents and two gender presentations, keeping linguistic content constant through voice cloning. Using pointwise LLM-judge ratings, pairwise comparisons, and Best-Worst Scaling with human validation, we detect recurring directional disparities. Eastern European-accented speech receives lower helpfulness scores, particularly for female-presenting voices. Responses remain polite but differ in helpfulness. While LLM judges capture the directional trend of these biases, human evaluators exhibit significantly higher sensitivity, showing stronger accent-level contrasts.

eess.AS↗

arXiv2026

Speak Your Mind: The Speech Continuation Task as a Probe of Voice-Based Model Bias

Speech Continuation (SC) is the task of generating a coherent extension of a spoken prompt while preserving both semantic context and speaker identity. Because SC is constrained to a single audio stream, it offers a more direct setting for probing biases in speech foundation models than dialogue does. In this work we present the first systematic evaluation of bias in SC, investigating how gender and phonation type (breathy, creaky, end-creak) affect continuation behaviour. We evaluate three recent models: SpiritLM (base and expressive), VAE-GSLM, and SpeechGPT across speaker similarity, voice quality preservation, and text-based bias metrics. Results show that while both speaker similarity and coherence remain a challenge, textual evaluations reveal significant model and gender interactions: once coherence is sufficiently high (for VAE-GSLM), gender effects emerge on text-metrics such as agency and sentence polarity. In addition, continuations revert toward modal phonation more strongly for female prompts than for male ones, revealing a systematic voice-quality bias. These findings highlight SC as a controlled probe of socially relevant representational biases in speech foundation models, and suggest that it will become an increasingly informative diagnostic as continuation quality improves.

eess.AS↗

arXiv2026

Gelina: Unified Speech and Gesture Synthesis via Interleaved Token Prediction

Human communication is multimodal, with speech and gestures tightly coupled, yet most computational methods for generating speech and gestures synthesize them sequentially, weakening synchrony and prosody alignment. We introduce Gelina, a unified framework that jointly synthesizes speech and co-speech gestures from text using interleaved token sequences in a discrete autoregressive backbone, with modality-specific decoders. Gelina supports multi-speaker and multi-style cloning and enables gesture-only synthesis from speech inputs. Subjective and objective evaluations demonstrate competitive speech quality and improved gesture generation over unimodal baselines.

cs.SD↗

arXiv2026

Towards Reliable Human Evaluations in Gesture Generation: Insights from a Community-Driven State-of-the-Art Benchmark

We review human evaluation practices in automatic, speech-driven 3D gesture generation and find a lack of standardisation and frequent use of flawed experimental setups. This leads to a situation where it is impossible to know how different methods compare, or what the state of the art is. In order to address common shortcomings of evaluation design, and to standardise future user studies in gesture-generation works, we introduce a detailed human evaluation protocol for the widely-used BEAT2 motion-capture dataset. Using this protocol, we conduct large-scale crowdsourced evaluation to rank six recent gesture-generation models -- each trained by its original authors -- across two key evaluation dimensions: motion realism and speech-gesture alignment. Our results show that 1) motion realism has become a saturated evaluation measure on the BEAT2 dataset, with older models performing on par with more recent approaches; 2) previous findings of high speech-gesture alignment do not hold up under rigorous evaluation, even for specialised models; and 3) the field must adopt disentangled assessments of motion quality and multimodal alignment for accurate benchmarking in order to make progress. To drive standardisation and enable new evaluation research, we release five hours of synthetic motion from the benchmarked models; over 750 rendered video stimuli from the user studies -- enabling new evaluations without requiring model reimplementation -- alongside our open-source rendering script, and 16,000 pairwise human preference votes collected for our benchmark.

cs.CV↗

arXiv2026

VoXtream2: Full-stream TTS with dynamic speaking rate control

Full-stream text-to-speech (TTS) for interactive systems must start speaking with minimal delay while remaining controllable as text arrives incrementally. We present VoXtream2, a zero-shot full-stream TTS model with dynamic speaking-rate control that can be updated mid-utterance on the fly. VoXtream2 combines a distribution matching mechanism over duration states with classifier-free guidance across conditioning signals to improve controllability and synthesis quality. Prompt-text masking enables textless audio prompting, removing the need for prompt transcription. Across standard zero-shot benchmarks and a dedicated speaking-rate test set, VoXtream2 achieves competitive objective and subjective results against public baselines despite a smaller model and less training data. In full-stream mode, it runs 4 times faster than real time with 74 ms first-packet latency on a consumer GPU.

eess.AS↗

arXiv2026

HiFi-Glot: High-Fidelity Neural Formant Synthesis with Differentiable Resonant Filters

Formant synthesis aims to generate speech with controllable formant structures, enabling precise control of vocal resonance and phonetic features. However, while existing formant synthesis approaches enable precise formant manipulation, they often yield an impoverished speech signal by failing to capture the complex co-occurring acoustic cues essential for naturalness. To address this issue, this letter presents HiFi-Glot, an end-to-end neural formant synthesis system that achieves both precise formant control and high-fidelity speech synthesis. Specifically, the proposed model adopts a source--filter architecture inspired by classical formant synthesis, where a neural vocoder generates the glottal excitation signal, and differentiable resonant filters model the formants to produce the speech waveform. Experiment results demonstrate that our proposed HiFi-Glot model can generate speech with higher perceptual quality and naturalness while exhibiting a more precise control over formant frequencies, outperforming industry-standard formant manipulation tools such as Praat. Code, checkpoints, and representative audio samples are available at https://www.yichenggu.com/HiFi-Glot/.

cs.SD↗

arXiv2026

Do Bias Benchmarks Generalise? Evidence from Voice-based Evaluation of Gender Bias in SpeechLLMs

Recent work in benchmarking bias and fairness in speech large language models (SpeechLLMs) has relied heavily on multiple-choice question answering (MCQA) formats. The model is tasked to choose between stereotypical, anti-stereotypical, or neutral/irrelevant answers given an input speech prompt and an optional text prompt. Such MCQA benchmarks implicitly assume that model performance is consistent across other MCQA tasks, voices, and other task formats such as more realistic, long-form evaluations. In this paper, we probe that assumption. We fine-tune three SpeechLLMs using LoRA adapters to induce specific MCQA behaviours: preference for stereotypical, anti-stereotypical, or neutral/uncertain answers. We then evaluate whether these behaviours generalise to another, distinct MCQA benchmark, and more critically to long-form, creative generation tasks. Our results show that performance on MCQA bias benchmarks fails to reliably predict performances across other MCQA benchmarks, and more importantly across long-form tasks. We conclude that current MCQA bias benchmarks show limited evidence of cross-task generalisation in the speech domain, and also propose an evaluation suite for measuring behaviour transferability in future models and benchmarks.

cs.CL↗

arXiv2026

VoXtream: Full-Stream Text-to-Speech with Extremely Low Latency

We present VoXtream, a fully autoregressive, zero-shot streaming text-to-speech (TTS) system for real-time use that begins speaking from the first word. VoXtream directly maps incoming phonemes to audio tokens using a monotonic alignment scheme and a limited look-ahead that does not delay onset. Built around an incremental phoneme transformer, a temporal transformer predicting semantic and duration tokens, and a depth transformer producing acoustic tokens, VoXtream achieves, to our knowledge, the lowest initial delay among publicly available streaming TTS: 102 ms on GPU. Despite being trained on a mid-scale 9k-hour corpus, it matches or surpasses larger baselines on several metrics, while delivering competitive quality in both output- and full-streaming settings. Demo and code are available at https://herimor.github.io/voxtream.

eess.AS↗

arXiv2025

A Non-Adversarial Approach to Idempotent Generative Modelling

Idempotent Generative Networks (IGNs) are deep generative models that also function as local data manifold projectors, mapping arbitrary inputs back onto the manifold. They are trained to act as identity operators on the data and as idempotent operators off the data manifold. However, IGNs suffer from mode collapse, mode dropping, and training instability due to their objectives, which contain adversarial components and can cause the model to cover the data manifold only partially -- an issue shared with generative adversarial networks. We introduce Non-Adversarial Idempotent Generative Networks (NAIGNs) to address these issues. Our loss function combines reconstruction with the non-adversarial generative objective of Implicit Maximum Likelihood Estimation (IMLE). This improves on IGN's ability to restore corrupted data and generate new samples that closely match the data distribution. We moreover demonstrate that NAIGNs implicitly learn the distance field to the data manifold, as well as an energy-based model.

cs.LG↗

arXiv2025

When Voice Matters: Evidence of Gender Disparity in Positional Bias of SpeechLLMs

The rapid development of SpeechLLM-based conversational AI systems has created a need for robustly benchmarking these efforts, including aspects of fairness and bias. At present, such benchmarks typically rely on multiple choice question answering (MCQA). In this paper, we present the first token-level probabilistic evaluation and response-based study of several issues affecting the use of MCQA in SpeechLLM benchmarking: 1) we examine how model temperature and prompt design affect gender and positional bias on an MCQA gender-bias benchmark; 2) we examine how these biases are affected by the gender of the input voice; and 3) we study to what extent observed trends carry over to a second gender-bias benchmark. Our results show that concerns about positional bias from the text domain are equally valid in the speech domain. We also find the effect to be stronger for female voices than for male voices. To our knowledge, this is the first study to isolate positional bias effects in SpeechLLM-based gender-bias benchmarks. We conclude that current MCQA benchmarks do not account for speech-based bias and alternative strategies are needed to ensure fairness towards all users.

eess.AS↗

arXiv2025

EmojiVoice: Towards long-term controllable expressivity in robot speech

Humans vary their expressivity when speaking for extended periods to maintain engagement with their listener. Although social robots tend to be deployed with ``expressive'' joyful voices, they lack this long-term variation found in human speech. Foundation model text-to-speech systems are beginning to mimic the expressivity in human speech, but they are difficult to deploy offline on robots. We present EmojiVoice, a free, customizable text-to-speech (TTS) toolkit that allows social roboticists to build temporally variable, expressive speech on social robots. We introduce emoji-prompting to allow fine-grained control of expressivity on a phase level and use the lightweight Matcha-TTS backbone to generate speech in real-time. We explore three case studies: (1) a scripted conversation with a robot assistant, (2) a storytelling robot, and (3) an autonomous speech-to-speech interactive agent. We found that using varied emoji prompting improved the perception and expressivity of speech over a long period in a storytelling task, but expressive voice was not preferred in the assistant use case.

cs.RO↗

arXiv2024

CARL-GT: Evaluating Causal Reasoning Capabilities of Large Language Models

Causal reasoning capabilities are essential for large language models (LLMs) in a wide range of applications, such as education and healthcare. But there is still a lack of benchmarks for a better understanding of such capabilities. Current LLM benchmarks are mainly based on conversational tasks, academic math tests, and coding tests. Such benchmarks evaluate LLMs in well-regularized settings, but they are limited in assessing the skills and abilities to solve real-world problems. In this work, we provide a benchmark, named by CARL-GT, which evaluates CAusal Reasoning capabilities of large Language models using Graphs and Tabular data. The benchmark has a diverse range of tasks for evaluating LLMs from causal graph reasoning, knowledge discovery, and decision-making aspects. In addition, effective zero-shot learning prompts are developed for the tasks. In our experiments, we leverage the benchmark for evaluating open-source LLMs and provide a detailed comparison of LLMs for causal reasoning abilities. We found that LLMs are still weak in casual reasoning, especially with tabular data to discover new insights. Furthermore, we investigate and discuss the relationships of different benchmark tasks by analyzing the performance of LLMs. The experimental results show that LLMs have different strength over different tasks and that their performance on tasks in different categories, i.e., causal graph reasoning, knowledge discovery, and decision-making, shows stronger correlation than tasks in the same category.

cs.CL↗

arXiv2024

Towards a GENEA Leaderboard -- an Extended, Living Benchmark for Evaluating and Advancing Conversational Motion Synthesis

Current evaluation practices in speech-driven gesture generation lack standardisation and focus on aspects that are easy to measure over aspects that actually matter. This leads to a situation where it is impossible to know what is the state of the art, or to know which method works better for which purpose when comparing two publications. In this position paper, we review and give details on issues with existing gesture-generation evaluation, and present a novel proposal for remedying them. Specifically, we announce an upcoming living leaderboard to benchmark progress in conversational motion synthesis. Unlike earlier gesture-generation challenges, the leaderboard will be updated with large-scale user studies of new gesture-generation systems multiple times per year, and systems on the leaderboard can be submitted to any publication venue that their authors prefer. By evolving the leaderboard evaluation data and tasks over time, the effort can keep driving progress towards the most important end goals identified by the community. We actively seek community involvement across the entire evaluation pipeline: from data and tasks for the evaluation, via tooling, to the systems evaluated. In other words, our proposal will not only make it easier for researchers to perform good evaluations, but their collective input and contributions will also help drive the future of gesture-generation research.

cs.HC↗

arXiv2024

Voice Conversion-based Privacy through Adversarial Information Hiding

Privacy-preserving voice conversion aims to remove only the attributes of speech audio that convey identity information, keeping other speech characteristics intact. This paper presents a mechanism for privacy-preserving voice conversion that allows controlling the leakage of identity-bearing information using adversarial information hiding. This enables a deliberate trade-off between maintaining source-speech characteristics and modification of speaker identity. As such, the approach improves on voice-conversion techniques like CycleGAN and StarGAN, which were not designed for privacy, meaning that converted speech may leak personal information in unpredictable ways. Our approach is also more flexible than ASR-TTS voice conversion pipelines, which by design discard all prosodic information linked to textual content. Evaluations show that the proposed system successfully modifies perceived speaker identity whilst well maintaining source lexical content.

cs.SD↗

arXiv2024

Should you use a probabilistic duration model in TTS? Probably! Especially for spontaneous speech

Converting input symbols to output audio in TTS requires modelling the durations of speech sounds. Leading non-autoregressive (NAR) TTS models treat duration modelling as a regression problem. The same utterance is then spoken with identical timings every time, unlike when a human speaks. Probabilistic models of duration have been proposed, but there is mixed evidence of their benefits. However, prior studies generally only consider speech read aloud, and ignore spontaneous speech, despite the latter being both a more common and a more variable mode of speaking. We compare the effect of conventional deterministic duration modelling to durations sampled from a powerful probabilistic model based on conditional flow matching (OT-CFM), in three different NAR TTS approaches: regression-based, deep generative, and end-to-end. Across four different corpora, stochastic duration modelling improves probabilistic NAR TTS approaches, especially for spontaneous speech. Please see https://shivammehta25.github.io/prob_dur/ for audio and resources.

eess.AS↗