NAMAA Saudi TTS V2

NAMAA Saudi TTS V2

Clones a voice from a short Arabic clip and its transcript, then reads any Arabic text you type in that voice, with Najdi dialect style. Non-commercial. Use only voices you have the right to use.

Made by xiiqj on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This is a Gradio demo of NAMAA-Saudi-TTS-V2, an F5-TTS model of about 335M parameters. Its training began with the Saudi checkpoint of Habibi-TTS, then added 18.4 hours of recordings in Najdi and Saudi Arabic. You give it a short clean clip and that clip's transcript, and it reads new Arabic text in the same voice.

Who it suits: Arabic content makers who have the right to use a voice, and researchers or developers who want to try dialect-aware voice cloning.

Industry | Who uses it | How they use it | Example
Media | Arabic video and audio creators | Read a script in a voice they have rights to | Voice a short narration from their own recording
Research | Speech researchers | Try Najdi-flavored voice cloning | Compare the output at different quality settings
Software | Developers | Call the synthesize function through the API or MCP | Add speech generation to an Arabic app
Education | Language teachers | Make spoken examples in a Najdi style | Record a sample sentence for a class

Skip it if you need commercial use, other Arabic dialects or a fixed built-in voice, since the license is non-commercial, the output leans Najdi and a reference clip is required.

Where to use it:

• A web browser, on the Space's page at Hugging Face
• Your own computer, as a local copy, once Python 3.12 is installed
• Python programs, through the Gradio client and the synthesize endpoint
• MCP clients, since the Space also exposes synthesize as an MCP tool

Use case | Who | What they get back
Clone a voice from a clip | Creators | Arabic speech in the voice of your clip
Tune speed and quality | Researchers | Speech made with your solver, speed and guidance settings
Call it from code | Developers | An audio result and a status from the API

How to set it up:

• Nothing at all if you only use the online Space
• Python 3.12 and pip, for a local copy
• The Python packages listed in requirements.txt
• A clean reference clip of 5 to 8 seconds, with its exact transcript

1. Open the Space's page at Hugging Face to use it online.
2. For a local copy, install the Hugging Face hub package and download the Space.
3. Move into the new folder and install Gradio at the pinned version.
4. Install the packages from requirements.txt.
5. Start the app with Python.

pip install -U huggingface_hub
hf download xiiqj/namaa-saudi-tts-v2 --repo-type space --local-dir namaa-saudi-tts-v2
cd namaa-saudi-tts-v2
pip install gradio==6.25.0
pip install -r requirements.txt
python app.py

The page names Habibi-TTS and a Vocos 24 kHz vocoder as parts of the setup. Use only voices you have the right to use.

How to use it: Upload or record a clean reference clip of 5 to 8 seconds with one speaker and full sentences. Type the exact transcript of that clip, because errors there hurt the output. Type the Arabic text to be spoken and press Generate speech.

Tool | What it does
Generate speech | Reads your text in the reference voice
nfe_step | Solver steps, with 16 for fast, 32 as default and 64 for quality
speed | Changes the speaking speed
cfg_strength | Changes how strongly the model follows the guidance
synthesize | The function exposed as a Gradio API endpoint and an MCP tool

From Python, the README shows a call through the Gradio client.

from gradio_client import Client, handle_file

c = Client("<namespace>/namaa-saudi-tts-v2")
audio, status = c.predict(
    ref_audio=handle_file("reference.wav"),
    ref_text="نص المرجع بالضبط",
    gen_text="النص المراد نطقه",
    api_name="/synthesize",
)

Example: You record a clear 6-second clip of your own voice reading a sentence, and type that sentence exactly as its transcript. Then you type a new Arabic line and press Generate speech. You get audio of the new line in your voice.

For newcomers and experts: If you are new, start with one of the default reference clips and a short text. The defaults are Najdi and Gulf samples, so you can hear the result before using your own recording.

If you know the field, this is a diffusion transformer fine-tuned from the Saudi checkpoint of Habibi-TTS. The recordings it learned from came from podcasts, so background room sound may leak into results. The advanced settings also cover chunk cross-fade and silence removal.

You add a 5 to 8 second reference clip
You type its exact transcript
You type the Arabic text to speak
The model synthesizes it in that voice
You get the audio

Good to know:

• License: cc-by-nc-sa-4
• Cost to run: none named, though the Space runs on zero-a10g hardware at Hugging Face
• Data it reaches: the reference clip and transcripts you give it, plus model files from Hugging Face
• Not for: commercial use, cloning a voice without permission, or deceptive or impersonating audio, and the README says it has no safeguards against misuse

On Hugging Face, 13 November 2026

• Likes: 1
• Built with: gradio
• License: cc-by-nc-sa-4.0

Links

• Space: https://huggingface.co/spaces/xiiqj/namaa-saudi-tts-v2
• Files: https://huggingface.co/spaces/xiiqj/namaa-saudi-tts-v2/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.