aiqbench analysis
This paper tackles a real limitation of alignment-free text-to-speech models: when they synthesize across languages, character-duration differences cause mismatches that hurt speech quality. Its main contributions are a cross-lingual benchmark covering English and nine Indian languages, and an inference-time duration estimator that uses language-specific per-character statistics, requiring no retraining or architectural changes. The reported gains are in naturalness, intelligibility, and speaker similarity on bilingual and multilingual A-DMA models, but readers should check how the benchmark and heuristics generalize beyond these models and languages.
Cross-Lingual TTS for Indian Languages: A Benchmark and Inference-Time Duration Estimation Strategies
arXiv:2610.14826v1 Announce Type: cross Abstract: Recent alignment-free Text-to-Speech (TTS) models achieve high synthesis quality by removing explicit alignment modules and simplifying training. However, they often struggle in cross-lingual scenarios due to differences in character durations across languages, leading to duration mismatch and degraded speech quality. Existing approaches typically address this issue using explicit duration predictors, increasing model complexity. Systematic evaluation of TTS performance for Indian languages in cross-lingual settings remains under-explored due to the lack of standardized evaluation benchmarks. In this work, we construct a cross-lingual evaluation benchmark covering English and nine Indian languages and propose a language-aware inference-time duration estimation strategy for alignment-free TTS that incorporates language-specific duration-per character statistics without retraining or architectural modifications. We further develop heuristic formulations that refine duration estimation based on language-dependent duration variations. Experiments on bilingual and multilingual A-DMA models demonstrate improved speech naturalness, intelligibility, and speaker similarity while reducing duration mismatch.
Abstract:Recent alignment-free Text-to-Speech (TTS) models achieve high synthesis quality by removing explicit alignment modules and simplifying training. However, they often struggle in cross-lingual scenarios due to differences in character durations across languages, leading to duration mismatch and degraded speech quality. Existing approaches typically address this issue using explicit duration predictors, increasing model complexity. Systematic evaluation of TTS performance for Indian languages in cross-lingual settings remains under-explored due to the lack of standardized evaluation benchmarks. In this work, we construct a cross-lingual evaluation benchmark covering English and nine Indian languages and propose a language-aware inference-time duration estimation strategy for alignment-free TTS that incorporates language-specific duration-per character statistics without retraining or architectural modifications. We further develop heuristic formulations that refine duration estimation based on language-dependent duration variations. Experiments on bilingual and multilingual A-DMA models demonstrate improved speech naturalness, intelligibility, and speaker similarity while reducing duration mismatch.
9 pages, 3 figures, 5 tables. Accepted at the AACL-IJCNLP 2026 Workshop Multi-LLL. The dataset and code recipes are available at: this https URL
Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Sound (cs.SD)
arXiv:2610.14826 [eess.AS]
(or arXiv:2610.14826v1 [eess.AS] for this version)
https://doi.org/10.48550/arXiv.2610.14826 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 269 words
Related coverage
LiveLaw Supreme Court Half-Yearly Digest: January - June 2026 [Part 2]
livelaw.in · Sep 13, 2026
Inference-Time Steering for Cross-Lingual Factual Consistency in LLMs
arXiv · Jul 23, 2026
Analyzing and Mitigating Cross-Lingual Degradation in Multilingual Medical VQA
arXiv · Aug 25, 2026
Cross-lingual Representation Learning via Centroid Intervention Fusion
arXiv · Aug 28, 2026