Whisper Valencian
BSC's Valencian Whisper release adapts OpenAI Whisper Large v3 to Valencian-accented Catalan speech.
Model and training
The card reports 397 hours and 55 minutes of adaptation data from Corts Valencianes and Valencian examples in Common Voice 21. The output is an unpunctuated Catalan transcript. The training data and accent focus distinguish these weights from the broad Catalan release. Its faster-whisper copy is a runtime conversion.Whisper Valencian: official model card
Practical use
Practical guidance. Use this model when Valencian speech is a major part of the input. Compare regional vocabulary and pronunciation against a broader Catalan recogniser. Parliamentary recordings and volunteer readings do not cover every conversation setting, so include informal speech and background noise in manual evaluation.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | whisper-3cat-cv21-valencianfaster-whisper-3cat-cv21-valencian |
| License | Apache License 2.0.Whisper Valencian: official model card |
Related models
- Whisper Balearic
Barcelona Supercomputing Center · BSC Whisper
BSC's Balearic Whisper release adapts OpenAI Whisper Large v3 to Balearic-accented Catalan.
Launch date not confirmedDownloadable BSC's Catalan Whisper model is a trained adaptation of OpenAI Whisper Large v3 for speech recognition.
Launch date not confirmedDownloadable