Features
Real-time streaming transcription with <300ms latency
Batch (async) transcription with no hallucinations
Partial transcripts in <100ms for real-time conversations
Solaria-3 model: 9.6% WER on real English audio
100+ languages with automatic language detection and code-switching
Speaker diarization (#1 on pyannoteAI benchmark)
PII redaction for sensitive data
Semantic sentiment analysis (94% confidence)
Named entity recognition (names, emails, addresses)
Chapterization and summarization
Audio-to-LLM pipeline (native or bring-your-own-model)
Translation and subtitles generation
Custom vocabulary and custom spelling
Adaptive handling of noisy, fast-paced, and conversational audio
100% EU data residency option
Automatic conversation recording via Pendant
Transcription of recorded conversations
Unlimited meeting notes (free for existing users)
Searchable transcript history
Data export tool for downloading transcripts
Wearable device (discontinued)
Desktop app recording (sunsetting)
Web app recording (sunsetting)
Rewind screen and audio capture (sunsetting)
Regional availability restrictions
