LEMAS: Large A 150K-Hour Large-scale Extensible Multilingual Audio Suite with Generative Speech Models

TL;DR

LEMAS is a comprehensive, large-scale multilingual speech dataset with 150,000 hours of annotated audio, enabling robust generative speech models for synthesis and editing across ten languages.

Contribution

The paper introduces the LEMAS-Dataset, the largest open-source multilingual speech corpus with word-level timestamps, and demonstrates its effectiveness through two novel benchmark models for speech synthesis and editing.

Findings

Models trained on LEMAS achieve high-quality multilingual synthesis.

The dataset enables effective speech editing with natural transitions.

Accent-adversarial training improves cross-lingual synthesis stability.

Abstract

We present the LEMAS-Dataset, which, to our knowledge, is currently the largest open-source multilingual speech corpus with word-level timestamps. Covering over 150,000 hours across 10 major languages, LEMAS-Dataset is constructed via a efficient data processing pipeline that ensures high-quality data and annotations. To validate the effectiveness of LEMAS-Dataset across diverse generative paradigms, we train two benchmark models with distinct architectures and task specializations on this dataset. LEMAS-TTS, built upon a non-autoregressive flow-matching framework, leverages the dataset's massive scale and linguistic diversity to achieve robust zero-shot multilingual synthesis. Our proposed accent-adversarial training and CTC loss mitigate cross-lingual accent issues, enhancing synthesis stability. Complementarily, LEMAS-Edit employs an autoregressive decoder-only architecture that formulates speech editing as a masked token infilling task. By exploiting precise word-level alignments to construct training masks and adopting adaptive decoding strategies, it achieves seamless, smooth-boundary speech editing with natural transitions. Experimental results demonstrate that models trained on LEMAS-Dataset deliver high-quality synthesis and editing performance, confirming the dataset's quality. We envision that this richly timestamp-annotated, fine-grained multilingual corpus will drive future advances in prompt-based speech generation systems.