Cohere announced the release of Cohere Transcribe Arabic, an open-source speech-to-text model designed for Arabic speech in business and developer contexts. The model is built on their 2B-parameter Automatic Speech Recognition (ASR) model and outperforms leading alternatives like Whisper v3 Large and OmniASR LLM 7B. It is available under the Apache 2.0 license and can be accessed through the Cohere API or Model Vault. Source: huggingface
Cohere Transcribe Arabic achieves the lowest average word error rate (WER) of any open-source model on the Hugging Face Arabic ASR Leaderboard, with a WER of 25.87. This represents a 2.45-point improvement over Meta’s OmniASR-LLM-7B and an 11-point improvement over OpenAI’s Whisper v3 Large. The model also delivers substantial gains over Cohere Transcribe on both Arabic and bilingual Arabic-English audio. Source: huggingface
The model was trained on a mix of Arabic, Arabic-accented English, and L1 English datasets, emphasizing dialect diversity, Arabic-English code-switching, and acoustic variety. The training data includes speech recorded across different devices, environments, and noise conditions, augmented with TTS-generated samples for underrepresented conditions. Source: huggingface
