# Choose a voice. Make your first narrated film.

The free Local Voice Add-on is included with the **$67 Make Video Skill toolkit**. Start with **Alba** at **1.25× pace**, or choose **Bill**, **Marius**, **Caro Davy**, **Stuart Bell** or **Javert**. Listen to the actual samples on the [setup page](../setup.html#included-voice), choose one and copy its request into your local toolkit chat. Your assistant runs the included local tools. The website does not install or run software on your computer.

## The first film

Keep the complete extracted `local-video-kit` and `Local-Voice-Add-on-1.3.0` folders together. Use Codex or Claude Code with permitted file and terminal access. The [voice-to-video request](setup-v13/prompts/new-local-voice.txt) checks the tools and selected voice setup, then makes one original nine-scene example with your chosen voice.

The complete production route is tested on Apple Silicon macOS with 32 GB memory. This is a tested environment, not a verified minimum or a complete-workflow guarantee for Windows, Linux or Intel Mac. Initial installation needs internet and compatible tools. Your assistant access and hardware remain separate.

For a manual run, replace the paths with the actual extracted folders:

```sh
python3 "ADDON/voice_video.py" doctor --kit "CORE"
python3 "ADDON/voice_video.py" make --kit "CORE" --voice alba --pace 1.25 --out "NEW-FILM-FOLDER" --install
```

Alba is the default. Other bridge IDs are `bill`, `marius`, `caro_davy`, `stuart_bell` and `javert`. `--out` must be a new folder. The default creates native full-HD wide and vertical MP4s. Keep the complete source, WAV files, SRT/VTT captions, credit sidecars and verification record.

The captions use estimated phrase timing derived from final paced scene audio. They are not forced-aligned word timings. Watch both movies with sound and review pronunciation, captions, crop and the final words. Technical checks do not replace listening.

## Make it about your own topic

Ask your assistant to copy the add-on's `first-film.story.json` into a new named story, write original narration and author meaningful new scenes. Each scene needs its own `narration` text. The bridge accepts 1–20 scenes, up to 600 narration characters per scene. Validate both layouts, then run:

```sh
python3 "ADDON/voice_video.py" make --kit "CORE" --voice alba --pace 1.25 --story "MY-STORY.json" --out "MY-NEW-FILM"
```

This is a native scene-plan workflow. It does not automatically turn arbitrary text into every toolkit style. Other styles retain their documented editable source and rendering tools. Existing films keep their approved narration unless you choose to change it.

## Add more voices when needed

The first selection contains **25 voices** for English scripts. Alba, Bill and Marius lead the selection; Caro Davy, Stuart Bell and Javert are three more featured choices. Some have different accents, so listen to your script before sharing.

An optional expansion adds **41 more voices for written scripts**, including the American voices below. Together that makes **66 standard voices**. There are also **13 advanced Japanese and Mandarin options** requiring specially prepared pronunciation input. They do not read ordinary written scripts through this first-video route. The add-on includes the full voice catalog, exact setup commands and source terms. Allow roughly 2 GB working space for the complete voice selection, plus project/video storage.

The larger roughly **6.7 GB Qwen planning, narration and forced-alignment stack** remains an [advanced optional route](free-local.md). It is not needed for your first film. Generate narration and render videos on your computer, with no per-video generation or rendering fees. AI assistant plans and optional paid media are separate. Hardware, electricity and downloads remain your costs.

## American voices

Hear the [American voice auditions](../voice-auditions-american.html): Heart, Bella, Sarah, Sky, Puck, Fenrir, Adam and Michael. These eight are part of the 66 standard voices. Listen for the delivery that fits your story.

American voice setup is optional and included at no extra charge with the $67 toolkit. Heart is the starting American voice. Choose a voice on the audition page and copy its request into your local assistant, or use the [Heart request](setup-v13/prompts/american-voice-video.txt). From the complete add-on folder, using the real extracted core path:

```sh
python3 voice_video.py doctor --engine kokoro --voice af_heart --kit "CORE"
python3 voice_video.py make --engine kokoro --voice af_heart --kit "CORE" --out "MY-NEW-HEART-FILM" --install
```

To choose another American narrator, select its name on the audition page or use its exact voice ID from the catalog with `--engine kokoro`. The first-video tools support all 66 standard voices. The 13 advanced Japanese and Mandarin options need the separate documented pronunciation workflow. Keep the generated credit records with your project and review the complete movie with sound. Alba remains the main starting voice; Heart is the starting American voice. Caption timing is estimated from the final narration.

## Credits and permissions

Keep the module's component licenses and generated source-credit records. Pocket code is MIT; its reviewed preset-only model/states use CC BY 4.0. Source recording terms vary by preset. Kokoro model/voices use Apache 2.0; its separate supplied worker and relevant phonemizer dependencies retain GPL terms. The add-on explains these independent rights. Do not imply performer endorsement or exclusive ownership of upstream voices.

The public audition files are existing local Pocket recordings of our original short script. [Audition credits](../media/local-voices/CREDITS.txt) record the exact sources and pitch-preserving 1.25× processing.