# Arabic TTS Inspection Kit v0.1

A small, free kit from Bayanat Labs for checking whether an Arabic text-to-speech voice, or a set of Arabic studio recordings, gets pronunciation and speaking style right, and for keeping each recording's voice rights in the record.

Page: https://www.bayanatlabs.com/research/arabic-tts-inspection-kit

## Files

| File | Contents |
|---|---|
| `rubric.md` | Listening protocol, 17 error codes, style scoring, how to report, and the rights check |
| `script.jsonl` | 18 Arabic test lines (14 MSA, 4 Saudi: 2 Najdi, 1 Hijazi, 1 Saudi code-switching) with plain text, a reference reading, tested words, error codes and what to listen for |
| `schema.json` | JSON Schema (draft 2020-12) for one studio utterance record: script, reading, style, audio, inspection result and separate rights fields |
| `manifest_example.jsonl` | 3 illustrative records (marked `"illustrative": true`): a number-gender failure, a Najdi pass whose synthetic-voice right is not stated, and a style mismatch |
| `SHA256SUMS.txt` | Checksums |
| `LICENSE.txt` | CC BY 4.0 |

## What it is not

- **No audio.** The kit is text only. It contains no recordings and no voice.
- **Not a sample of a Bayanat dataset.** The lines were written by Bayanat Labs for this kit. The speakers, sessions, inspectors and consent IDs in `manifest_example.jsonl` are fictional.
- **Not a benchmark.** Eighteen lines are a smoke test. Extend the script with your own domain terms before launch.
- **Not a rights grant.** The rights fields show how to record permissions; they do not grant any.

## Cite

Bayanat Labs (2026). Arabic TTS Inspection Kit v0.1. https://www.bayanatlabs.com/research/arabic-tts-inspection-kit

Licensed CC BY 4.0. Corrections and requests: hello@bayanatlabs.com