Call Transcription
Every call is automatically transcribed with speaker diarization (identifying who said what). Powered by ElevenLabs Scribe for accuracy and speed.
How transcription works
When a call ends, Coldread downloads the recording from your VoIP provider and sends it to ElevenLabs Scribe for transcription. The transcript includes:
- Full text of the conversation
- Speaker labels (Speaker 1, Speaker 2, etc.)
- Timestamps for each utterance
- Confidence scores
Processing time
Most calls are transcribed within 30-60 seconds of completion. Longer calls (10+ minutes) may take up to 2 minutes.
Accuracy
ElevenLabs Scribe provides industry-leading accuracy for UK, US, and Australian accents. Speaker diarization correctly identifies speakers in 95%+ of calls.
Supported languages
Currently supported:
- English (UK, US, AU, CA)
Additional languages coming soon.
Related
- Core concepts — how transcription fits into the processing pipeline
- Aircall integration — connect Aircall for automatic transcription
- Ringover integration — connect Ringover for automatic transcription
- Compliance monitoring — run compliance checks on transcripts
- Conversation intelligence vs call recording — why transcription is just the start