Enterprise AI innovator Cohere has officially unveiled its inaugural voice model, Transcribe, an open-source automatic speech recognition (ASR) solution designed for diverse applications such as comprehensive note-taking and sophisticated speech analysis.
Notably lightweight with just 2 billion parameters, Transcribe is engineered for seamless deployment on consumer-grade GPUs, catering to users preferring to self-host the model. It currently offers support for 14 languages, including English, French, German, Italian, Spanish, Portuguese, Greek, Dutch, Polish, Chinese, Japanese, Korean, Vietnamese, and Arabic.
Cohere asserts that Transcribe outperforms several prominent models on the Hugging Face Open ASR leaderboard, including Zoom Scribe v1, IBM Granite 4.0 1B, ElevenLabs Scribe v2, and Qwen3-ASR-1.7B Speech. It achieved an impressive average word error rate (WER) of 5.42, marking it as the lowest on the benchmark.
Furthermore, the company highlights that Transcribe secured an average win rate of 61% against competing models in evaluations where human assessors judged transcriptions for accuracy, coherence, and usability. However, it did exhibit some limitations, underperforming its rivals when transcribing Portuguese, German, and Spanish.
Transcribe also boasts remarkable efficiency, capable of processing 525 minutes of audio within a single minute, a speed considered exceptionally high for a model in its category, according to Cohere.
Cohere plans to integrate Transcribe into its enterprise agent orchestration platform, North, and is making the model freely accessible via its API. Additionally, Transcribe will be available on Model Valut, Cohere’s managed inference platform, broadening its reach and utility.
The launch of Transcribe comes at a time of escalating demand for speech recognition models, fueled by the growing popularity of advanced note-taking and dictation applications like Granola and Wispr Flow.
In related company news earlier this year, Cohere reportedly informed investors of a projected annual recurring revenue of $240 million by 2025. CEO Aidan Gomez was also quoted, indicating the startup “may go public soon.”
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.