OpenAI Whisper
Robust speech recognition via large-scale weak supervision. SOTA multilingual STT.
81.5kstars
speech-to-text
Advantages
- +Top-tier multilingual recognition accuracy
- +Strong noise robustness
- +Multiple model sizes for different hardware tiers
Limitations
- -Time drift on long audio transcriptions
- -Poor domain-specific terminology recognition
- -No real-time streaming transcription