Cohere has introduced Transcribe, its first open-source automatic speech recognition model, marking an expansion into AI voice technology for enterprise and developer use. Designed for tasks such as note-taking and speech analysis, the model is optimized for deployment on consumer-grade GPUs and supports 14 languages, including English, Chinese, Arabic, and Japanese.
The 2-billion-parameter model delivers competitive performance, achieving a leading word error rate on industry benchmarks and demonstrating strong accuracy and usability in human evaluations. Cohere also reported high processing efficiency, enabling rapid transcription of large audio volumes.
Transcribe will be integrated into Cohere’s North platform for agent orchestration and made available via API and its Model Vault infrastructure, as the company strengthens its position in enterprise AI tools.
