Voiceover

Overview

Each dialog block can have a voiceover — AI-generated speech from the character's assigned voice.

Requirements

• The character assigned to the block must have a Voice ID configured (ElevenLabs voice ID)
• The user must have voiceover:generate permission

Workflow

1. Hover over a dialog block → click the microphone (🔊) button
2. VoiceMasterModal opens — shows the block's text pre-filled
3. Edit the text if needed (e.g. adjust pronunciation)
4. Click Generate → ElevenLabs TTS generates the audio
5. Audio auto-plays for preview
6. Click Accept to upload and save the voiceover to the block
   — or —
   Click Regenerate to try again
7. The block now shows a ring on the voice button indicating a saved voiceover

Listening to a saved voiceover

The Listen button in the modal fetches the audio via apiFetch (with Bearer token) and plays it from a blob URL. The audio file is served by the backend API — direct <audio src> links won't work because they bypass auth headers.

Downloading

Click the download button to save the MP3 file locally. This also uses apiFetch to fetch with auth, then triggers a browser download.

Deleting

Click the trash icon (requires voiceover:delete). This deletes the stored audio file from the server and clears the voiceUrl on the block.

Inline playback (scene view)

The play button on each dialog block in the scene view also plays the voiceover. It uses the same fetch-to-blob pattern.

API

Method	Endpoint	Description
POST	/api/voices/:voiceId/generate	Generate speech (returns audio blob)
POST	/api/voices/upload	Upload audio blob, returns stored URL
DELETE	/api/voices/file	Delete a voice file by URL