LEMAS publishes a 150,000-hour speech dataset
A January 12, 2026 publication presents an audio dataset with 150,000 hours across 10 languages, word-level timestamps, and the generative models LEMAS-TTS and LEMAS-Edit.
Source: LEMAS lança dataset multilíngue de speech com 150 mil horas e modelos (lemas-project.github.io). Text prepared with AI from this source.
What happened and what to do
In a January 12, 2026 publication, LEMAS presented a multilingual audio dataset with 150,000 hours, coverage of 10 languages, and word-level timestamps. The publication also describes the generative models LEMAS-TTS and LEMAS-Edit for speech generation and editing.
A company evaluating voice applications can start by mapping languages, use cases, and data rights and quality requirements. Based on that diagnosis, it can implement an audio ingestion and validation pipeline, test speech generation and editing, and track results in a dashboard, without assuming the dataset suits every product or context.
How the consultancy can help
Wendelmaques can diagnose the opportunity and data, quality, and operational requirements, define a scope, and implement and operate an audio pipeline, speech tests, and monitoring tailored to the company’s case.
Next step
Send a short description of your voice use case to receive a proposal scoped for diagnosis and implementation.
Let's assess your project
I reply with scope and a proposal.