How to put multilingual
character voices into your game.

For creators and for developers

Five steps, start to finish

The important link is line_id. Keep it the same across your script, translation files and game events so audio never gets assigned to the wrong line.

Start with the dialogue you already have

Example: you write a scene with Maya and a guard. Your game exports a CSV with a stable ID, character, line, and scene context. SceneVoice would use the context to help create suitable localized dialogue.

line_id,character,scene,text,delivery
GATE_014,Maya,North Gate,"Open the gate!",urgent
GATE_015,Guard,North Gate,"They're coming!",afraid

What the studio does: export dialogue from its writing tools. No need to manually rename thousands of recordings.

Choose languages and request recordings

In the web workspace, this would be a simple form. A future API integration would let your build pipeline send the same information automatically. Authorized voices are associated with characters in your project.

curl -X POST https://api.scenevoice.example/v1/jobs \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "project_id": "castle_game",
    "script_revision": "v2",
    "languages": ["es", "fr"],
    "lines": [{
      "line_id": "GATE_014",
      "character_id": "maya",
      "text": "Open the gate!",
      "context": "urgent warning"
    }]
  }'

{"job_id":"job_example_104","status":"queued"}

The job ID would let the studio retrieve results later rather than wait for a long audio-generation request.

Listen, correct, approve

Translation is not automatically final. A reviewer checks meaning and game terminology; the audio director checks voice rights, emotion, pronunciation and timing. Only approved takes are released in the export package.

A proposed job-status endpoint could report queued, processing, review_required, completed, or failed. Future signed webhooks could notify the studio of status changes.

Download game-ready audio, not a pile of mystery files

The planned export package includes approved WAV recordings plus a machine-readable manifest, so engine import tools can match each asset to its dialogue line.

scenevoice-export-v2/
  audio/
    es/GATE_014.wav
    fr/GATE_014.wav
  dialogue.csv
  manifest.json

{
  "project": "castle_game",
  "revision": "v2",
  "assets": [{
    "line_id": "GATE_014",
    "language": "es",
    "character": "maya",
    "audio_file": "audio/es/GATE_014.wav",
    "approval": "approved"
  }]
}

How a developer would use it in Unity or Unreal

These are suggested ways to import exported assets using existing Unity or Unreal tools; no dedicated engine plugin is required by the proposed API workflow.

Change one sentence. Keep the rest.

Because the line_id stays stable, SceneVoice is designed to flag Spanish and French takes of GATE_014 for retranslation and reapproval. Unchanged, approved lines can be reused. The updated manifest lets the developer replace the affected assets in the game.

The endpoints we intend to build

Proposed security: project-scoped bearer keys, HTTPS, authorized voice usage, signed webhook events and permission-controlled exports. Endpoint paths and schemas are preliminary and may change.