Cohere introduces open-source voice model built for transcription

Enterprise AI company Cohere on Thursday introduced its first voice model, called Transcribe, an open-source automatic speech recognition system designed for applications such as note-taking and speech analysis.

With a relatively lightweight design at just 2 billion parameters, the model is intended to run on consumer-grade GPUs, making it suitable for users who prefer to self-host. It currently supports 14 languages, including English, French, German, Italian, Spanish, Portuguese, Greek, Dutch, Polish, Chinese, Japanese, Korean, Vietnamese, and Arabic.

Cohere stated that Transcribe outperforms several competing models — including Zoom Scribe v1, IBM Granite 4.01 B, ElevenLabs Scribe v2, and Qwen3-ASR-1.7B Speech — on the Hugging Face Open ASR leaderboard. The model reportedly achieved an average word error rate (WER) of 5.42, which is lower than any other model currently listed on the benchmark.

According to the company, Transcribe also demonstrated an average win rate of 61% when human evaluators compared its output against other models, assessing factors such as transcription accuracy, coherence, and usability. However, its performance lagged behind competitors in certain languages, particularly Portuguese, German, and Spanish.

Cohere added that the model can process 525 minutes of audio in just 1 minute, a speed that stands out in its category.

The company plans to integrate Transcribe into its enterprise agent orchestration platform, North, and to offer access to the model through its API at no cost. In addition, it will be available on Model Vault, Cohere’s managed inference platform.

Speech recognition technology has been gaining traction as demand for tools that support note-taking and dictation increases, including apps like Granola and Wispr Flow.

Earlier this year, Cohere reportedly informed investors that it was generating annual recurring revenue of $240 million in 2025. Its CEO, Aidan Gomez, was also quoted as suggesting that the company could pursue a public listing in the near future.