Whisper-large-v3

Whisper-large-v3 is OpenAI's state-of-the-art automatic speech recognition (ASR) model, delivering best-in-class accuracy across 99+ languages.

This is the gold standard for high-accuracy speech recognition (ASR) and translation (speech-to-English). Built on OpenAI's transformer-based architecture, the model features 1.55 billion parameters and handles challenging audio (noise, accents, technical terms) for over 99 languages. It delivers a significant performance upgrade: specifically, expect a 10% to 20% error reduction compared to the previous large-v2 version. Use it for precision tasks: legal transcription, academic research, and multilingual content creation demand this level of reliability.

https://huggingface.co/openai/whisper-large-v3