Speaker selection for interviews

Pick your voice on multi-speaker recordings so posts use only your lines.

When you create a project with Host, Guest, or Advisor perspective on a multi-speaker recording, Sparky diarizes the audio after transcription and pauses at awaiting_speaker status.

How to pick your voice

  1. Open the project page when status is awaiting_speaker (you also get an in-app notification).
  2. Review 2-3 sample lines from each detected speaker.
  3. Select the voice that matches you on the mic.
  4. Processing continues - only your utterances feed moment extraction and post generation.

Transcript file uploads (.txt / .srt) skip speaker selection and only support Full Conversation (podcast) or Trainer (Knowledge & Advisory). Use URL or audio upload when you need Host, Guest, or Advisor speaker selection.

Host-led podcasts that use the full episode should use Full Conversation perspective. Use Host when you want only your lines, not the guest's.