We firmly believe that the best way to learn a language is through extensive exposure to "comprehensible input," the famous "i+1" theory. This means content should be slightly above your current level ...
Cohere has released Transcribe, a 2-billion-parameter open-source speech recognition model that tops the Hugging Face Open ...
Mistral AI launches Voxtral TTS, an open-weight enterprise voice model that runs on a smartphone and challenges ElevenLabs in ...
Relatively light at just 2 billion parameters, the model is meant for use with consumer-grade GPUs for those who want to self ...
Mistral releases Voxtral TTS model that’s fast, multilingual and small enough to be practical for voice agents.