tech
March 26, 2026
Cohere launches an open-source voice model specifically for transcription
Relatively light at just 2 billion parameters, the model is meant for use with consumer-grade GPUs for those who want to self-host it. It currently supports 14 languages.
TL;DR
- Cohere has launched an open-source voice model for transcription.
- The model has 2 billion parameters, making it relatively light.
- It is intended for use with consumer-grade GPUs, enabling self-hosting.
- The model currently supports 14 languages.
Continue reading the original article