OpenAI's Whisper is an open-source speech recognition model with near-human accuracy across 99 languages. Available via API and local installation for transcription applications.
