# Arabic voice acting with AI for games and animation: characters that shout, tremble and laugh

> A game character who sounds calm while a building collapses breaks the scene. nutq-speak-3 gives Arabic characters the range they need, angry, scared, surprised, laughing, from cues written into each line.

- Source: https://nutq.dev/blog/arabic-ai-voice-acting-games-animation
- Published: 2026-10-25
- Updated: 2026-10-25
- Author: nutq team (https://nutq.dev)

## Short answer

For Arabic voice acting with AI, cast each character from the 16 Gulf voices or clone an actor with their consent, write every line with an emotion cue such as [angry], [scared], [surprised] or [laughter], and generate one file per line ID through the API. Give short barks a few words, budget about 13 characters per second for timing, and expect around a dollar for 300 lines.

A hero who sounds calm while the building comes down breaks the scene. **Arabic voice acting** with AI has mostly meant one polite reading per line, which works for a tutorial prompt and fails for a fight, a chase or a joke. nutq-speak-3 gives Arabic characters the range a script needs: angry, scared, surprised, tender, laughing, from cues you write into each line, in Gulf voices that sound like the region the game is set in.

## What can AI voice acting do for an Arabic game or cartoon?

| Use | What it replaces | Why it helps |
|---|---|---|
| Scratch dialogue for animatics | Developers reading lines themselves | Timing and tone are real from day one |
| Secondary characters and crowds | Booking many actors for a few lines each | Dozens of distinct voices on demand |
| Localising an English game into Gulf Arabic | A full dubbing session per patch | New lines in the same voices for every update |
| Kids' cartoons and explainers | Flat narration | Characters that laugh and react |
| Barks and combat lines | Silence, or English only | Short reactions in every state |

Many studios keep human actors for the lead roles and use AI for the rest; smaller teams ship whole games with it. Both work, because the voices stay the same from the first build to the last patch.

## How do you cast the characters?

Start from the 16 Gulf voices and audition each candidate with the same three lines: one calm, one angry, one laughing. Some starting points:

| Character | Try | Notes |
|---|---|---|
| Hero | Sultan, Hamdan | Holds up in shouting and in quiet scenes |
| Heroine | Maitha, Alia | Natural in banter and fear |
| Mentor or elder | Salem, Hamad | Steady, carries narration too |
| Comic sidekick | Abdullah, Juma | Conversational, laughs easily |
| Villain | Mansour, Saif | Weight and edge |
| Narrator | Hind | Even and warm over long reads |

Hear them in the [voice library](/voices). For a signature character, you can clone a voice actor from six to eight seconds of recording, with their written consent, and direct the clone with the same cues. See [Arabic text to speech with voice cloning](/blog/arabic-text-to-speech-voice-cloning) for what makes a good reference.

## How do you direct emotion line by line?

Put one cue where the feeling starts. A scene with three characters:

```text
[Sultan] [scared] سمعتوا هالصوت؟ [pause:700] الأرض قاعدة تهتز!
[Maitha] [angry] قلت لك لا تلمس الباب! [pause:400] شو سويت؟
[Abdullah] [amused] أنا؟ [laughter] أبد، الباب فتح بروحه.
[Salem] [tender] هدّوا… [pause:600] [neutral] نطلع من هني، وبسرعة.
```

Game states map cleanly to cues: `[excited]` and `[laughter]` for a win, `[sad]` and `[sigh]` for a defeat, `[surprised]` for a discovery, `[angry]` for combat taunts. If a performance is too strong, move the cue later in the line, or drop it and let the words carry it. Never stack several cues on one sentence; it sounds acted. The full cue list is in [Arabic text to speech with emotion](/blog/arabic-text-to-speech-with-emotion), and for children's characters, see [Arabic bedtime stories for kids](/blog/arabic-kids-stories-ai-voice).

## How do you handle barks and very short lines?

Very short lines, one or two words, give any expressive model the least to work with, so they are the lines to check first. Give each bark a few words of context: «وقف مكانك! لا تتحرك!» performs better than «وقف!». Write three or four wordings of each bark rather than generating the same text again, because nutq-speak-3 returns the same reading for the same text and voice. Different wordings also stop the player hearing the same line a hundred times.

## How do you get the lines into the engine?

Keep the script as a table of line IDs, character, emotion and text, and generate one WAV per ID through the API so the files drop straight into your engine's dialogue system. Name each file after its line ID, such as `ch02_sultan_017.wav`. For timing and lip-sync budgets, about 13 characters of Arabic make one second of speech. English names inside a line, such as a weapon or a place, are read as English. Before a long batch, run the free script check in the [MCP tools](/app/mcp): it lists every speaker and emotion it found and flags cues it would skip. Parameters and samples are in the [API docs](/docs).

When localising from English, write the Arabic line for the moment, not word for word, then check its length against the animation. A literal translation is usually longer and stiffer than what a Gulf speaker would say.

## What does it cost?

$0.05 per 1,000 characters. A game with 300 lines of about 60 characters each is 18,000 characters, roughly 90 cents, and pauses are free. Shipping a game or a show is commercial use, which the monthly plans license. To hear a character before you commit, type a line on the [Arabic AI voice generator page](/arabic-text-to-speech).

## FAQ

### Can AI do Arabic voice acting for games?

Yes. nutq-speak-3 performs Gulf Arabic lines with seven emotions plus real laughter and sighs, in 16 voices. Studios use it for scratch dialogue and secondary characters, and smaller teams ship whole games with it.

### Can I clone a voice actor's voice?

Yes, with the actor's written consent. A clone needs six to eight seconds of clean recording, and the cloned voice takes the same emotion cues as the library voices.

### Does nutq lip-sync the animation?

No. nutq produces the audio. Your animation or game tool handles lip-sync; for planning, about 13 characters of Arabic make one second of speech.

### How do I get a different take of a line?

Change the line slightly. nutq-speak-3 gives the same reading for the same text and voice, so move the cue, add a comma or rephrase a word to get a new performance.

### What does voicing a game cost?

Speech is $0.05 per 1,000 characters. A game with 300 lines of about 60 characters each is 18,000 characters, roughly 90 cents. Shipping it commercially needs a monthly plan, which includes the commercial licence.