ConMamba Small Spanish
ConMamba Small Spanish is BSC's Spanish speech-recognition release using a convolution and state-space architecture.
Model and training
The card reports training on 2,788 hours of Spanish television and radio material. It generates unpunctuated text. It is a separately trained language checkpoint, not a Spanish setting for the Catalan model. GPL 3.0 applies to this release.ConMamba Small Spanish: official model card
Practical use
Practical guidance. Use speech beyond broadcast recordings when assessing suitability for meetings or voice notes. Microphone conditions, interruptions and spontaneous wording can differ substantially from the training source. Compare recognition quality and resource use with the multilingual LoS model before choosing a dedicated Spanish deployment.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | ConMamba-small-es |
| License | GNU General Public License 3.0.ConMamba Small Spanish: official model card |
Related models
- ConMamba Small Catalan
Barcelona Supercomputing Center · BSC ConMamba
ConMamba Small Catalan is BSC's Catalan speech recogniser built with convolution-augmented Mamba state-space layers.
Launch date not confirmedDownloadable Whisper Large v3 LoS is BSC's multilingual adaptation for Spanish, Catalan, Galician and Basque speech.
Launch date not confirmedDownloadable