Matxa-TTS v2 Multi-Accent Catalan
Matxa-TTS v2 Multi-Accent Catalan generates speech from graphemes with multiple Catalan accents and speakers.
Model and training
BSC trains this Matcha-TTS-based flow-matching model from scratch on LaFresCat, ReFresCat and EnVeuAlta studio recordings. It can use the AlVoCat vocoder. The official ONNX package joins the acoustic model with a WaveNeXt vocoder; that package is not a new trained model. Commercial use requires a separate arrangement with the voice artists.Matxa-TTS v2 Multi-Accent Catalan: official model card
Practical use
Practical guidance. Evaluate the requested accent and voice explicitly, rather than treating Catalan coverage as one uniform result. Check names and words that distinguish accents. Keep the permitted use of the voice separate from the technical ability to generate it, and verify the custom terms before using the model in a paid service.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | matxa-tts-v2-ca-multiaccent-graphemes |
| License | Custom research/non-commercial OpenRAIL-M terms; commercial licensing requires the voice artists' agreement.Matxa multi-accent: custom model and voice license |
Related models
Matxa-TTS v2 Central Catalan is BSC's grapheme-based speech model built with the Matcha-TTS flow-matching approach.
Launch date not confirmedDownloadable- StyleTTS2 Catalan Multispeaker
Barcelona Supercomputing Center · BSC StyleTTS2
BSC's Catalan StyleTTS2 release trains a multispeaker speech synthesiser for Catalan.
Launch date not confirmedDownloadable