speech.transcribe

To transcribe audio with speaker labels, request diarization and word or segment timestamps. The Skill processes the audio file and returns structured transcript artifacts with speaker-attributed segments and metadata.

Yes, speech transcription supports attachments by filename, file path, cached media references, and HTTPS URLs. This allows processing audio from remote sources without requiring local file uploads.

The best way to get a compliance call recap is to transcribe the uploaded audio with diarization and word timestamps enabled. This generates speaker-attributed segments and word-level timing for detailed review.

Yes, provider-agnostic speech transcription supports auto selection or manual override between OpenAI, Deepgram, and AssemblyAI. This ensures practical reliability across different provider-ready scenarios.

Language detection automatically identifies the spoken language during audio transcription. The detected language is included in the structured output metadata alongside the transcript text and segment data.

Diarization produces speaker-attributed segments only when explicitly enabled in the transcription request. Without enabling it, the output provides structured text and timestamps without distinct speaker labeling.