Cohere releases Transcribe Arabic, an open-source model for Arabic transcription

Cohere releases *Transcribe Arabic*, a 2-billion-parameter open-source model for Arabic speech recognition, outperforming Whisper and OmniASR on dialects and Arabic-English code-switching.

Published 1sem1 sourceNotable

Lire en français

≈ 24sSpeed

The fact

The model is tailored for Arabic's unique challenges, including dialectal variations and bilingual speech.

Available on Hugging Face under Apache 2.0 license, it aims to enhance transcription accessibility for Arabic.

Click the link to read an article on the topic:

Dev.to

Explore this topic

Cohere #transcription #reconnaissance vocale #arabe #dialectes #code-switching #open-source #Apache 2.0 #Hugging Face

What if you saw the whole news differently?Factae cross-checks hundreds of sources worldwide to keep only the fact, no opinion. Explore the front page.

Auto-synthesis from 1 media source · identified on July 7, 2026

← Back to home

The fact

Read more

Anthropic unveils tool to decipher its AI Claude's 'thought process'

Kimi K3: China unveils 3-trillion-parameter open-source AI

Google DeepMind and Isomorphic Labs unveil their AI-driven bioresilience approach

Apple secures approval to deploy Apple Intelligence in China

Kimi K3: A new conversational AI model launched

EU rejects OpenAI's trademark application for « OPENAI »