Cohere has released an open-source voice model designed specifically for transcription tasks. The model is relatively lightweight at 2 billion parameters, enabling it to run on consumer-grade GPUs for self-hosting. It currently supports transcription in 14 languages.
Background
Cohere is a prominent AI company known for its large language models and enterprise-focused AI solutions. The field of speech-to-text transcription has seen increasing demand for efficient, accessible, and customizable models.
- Source
- TechCrunch
- Published
- Mar 26, 2026 at 09:30 PM
- Score
- 6.0 / 10