Audio

#Choose a voice

Browse voices opens Choose a voice with five top-level tabs: Favorites, ElevenLabs, MiniMax, Fish Audio, and AI33 Pro (which reveals nested ElevenLabs, MiniMax, and Fish Audio routes powered by AI33 Pro). Search by name or voice ID and use the filters or model selector each route offers; more voices load as you scroll. Play a preview before choosing.

• Star a voice to add it to Favorites. Favorites remember the exact route through which the voice was saved (Direct vs. AI33 Pro).
• Automatic pipeline narration uses the voice that has been selected in Audio Studio (whether from a Direct route — ElevenLabs, MiniMax, Fish Audio — or from AI33 Pro). Clicking Use or the robot button on any voice makes it the pipeline voice. The Audio page shows Pipeline voice: …; with none selected, automatic narration waits.
• Use selects the voice for both Audio Studio narration and automatic pipeline runs.
• If a provider route's voices cannot be loaded (for example its API key is missing), the reason is shown in the voice list. Configure or test your keys in AI providers.

#Generate the narration

1. Choose the project and approved script; the Narration editor loads the script text.
2. Edit the text if needed (chapter titles are spoken, except the opening hook title).
3. Click Generate audio. The job appears in the Audio queue and finishes in the background.

Each narration also asks the voice provider for a transcript with timings. Flow saves it with the audio, so the editor timeline cuts each scene where it is spoken and the captions match the speech.

#Audio queue and history

The Audio queue shows running and recent narrations; play or cancel them there. History keeps every completed, active and failed narration of your account.