Cohere Introduces Open-Source Voice Model for Transcription

Cohere has announced the release of an open-source voice model specifically designed for transcription. With only 2 billion parameters, the model is intended to run on consumer-grade GPUs, making self-hosted deployment more accessible.

The model currently supports 14 languages, expanding the possibilities of use in different geographical and linguistic contexts. The choice of a relatively lightweight model in terms of parameters suggests a focus on efficiency and reducing computing costs, a crucial aspect for those wishing to manage inference on-premise.

For those evaluating on-premise deployments, there are trade-offs to consider carefully. AI-RADAR offers analytical frameworks on /llm-onpremise to evaluate these aspects.