Cohere releases an open source audio template specifically for transcription

AI company Cohere on Thursday launched its first audio model: Transcribe, an open source automatic speech recognition model that can be used for tasks like note-taking and speech analysis.

Relatively lightweight at just 2 billion parameters, the model is intended for use with consumer GPUs for those who want to self-host it. It currently supports 14 languages: English, French, German, Italian, Spanish, Portuguese, Greek, Dutch, Polish, Chinese, Japanese, Korean, Vietnamese, and Arabic.

Cohere says Transcribe outperforms models like Zoom Scribe v1, IBM Granite 4.0 1B, ElevenLabs Scribe v2, and Qwen3-ASR-1.7B Speech on the Hugging Face Open ASR leaderboard, achieving an average word error rate (WER) of 5.42, which is lower than any other model in the benchmark.

The company claims that Transcribe achieved an average win rate of 61% compared to other models when human raters evaluated its transcripts for accuracy, consistency, and ease of use. However, this model fell behind its competitors when it had to copy Portuguese, German and Spanish.

Cohere says the Transcribe can process 525 minutes of audio in one minute, which is high for its model class.

The company plans to integrate Transcribe into its enterprise agent orchestration platform, North, and make the model available through its API for free. The model will also be available on Model Vault, Cohere’s managed inference platform.

Speech recognition models are growing in popularity as demand for note-taking and dictation apps like Granola and Wispr Flow grows.

Earlier this year, Cohere reportedly told investors that it would reach $240 million in annual recurring revenue in 2025, and its CEO, Aidan Gomez, was quoted as saying that the startup could go public “soon.”