Cohere launches an open-source voice model specifically for transcription

Cohere, a leader in enterprise AI, has made waves by introducing its first open-source voice model, named Transcribe. This innovative automatic speech recognition model is designed for tasks such as note-taking and speech analysis, making it a valuable tool for a variety of users. Transcribe is impressively lightweight, featuring 2 billion parameters, enabling it to operate efficiently on consumer-grade GPUs for those who wish to self-host the solution. The model currently supports a robust array of 14 languages, including English, French, German, Italian, Spanish, Portuguese, Greek, Dutch, Polish, Chinese, Japanese, Korean, Vietnamese, and Arabic. Cohere proudly claims that Transcribe outperforms several prominent models, including Zoom Scribe v1 and IBM Granite 4.0 1B, as evidenced by its top position on the Hugging Face Open ASR leaderboard. It boasts an average word error rate (WER) of just 5.42, surpassing all competitors on the benchmark. In evaluations conducted by human testers, Transcribe achieved an impressive average win rate of 61% when it came to transcription accuracy, coherence, and usability. Despite its accolades, Transcribe faced challenges with transcribing Portuguese, German, and Spanish, where it did not perform as well as expected. Nonetheless, it can process an impressive 525 minutes of audio in a single minute, a significant feat for models in its category. Looking ahead, Cohere plans to integrate Transcribe into its enterprise agent orchestration platform, North. The model will also be available through its API at no cost and can be accessed via Model Vault, Cohere’s managed inference platform. The rising popularity of speech recognition models is evident, as demand surges for note-taking and dictation applications like Granola and Wispr Flow. Earlier this year, Cohere indicated to investors that it is on track for an annual recurring revenue of $240 million by 2025, with CEO Aidan Gomez hinting at a potential public offering in the near future.