Audar-TTS-V1-Turbo

<div align="center">

Audar-TTS-V1-Turbo · GGUF

Open, Arabic-first, expressive zero-shot text-to-speech — the balanced production default.

From Arabic to the world.

<p><a href="#-voice-gallery"><b>🎧 Voice Gallery</b></a> · <a href="#local-deployment-gguf"><b>💻 Local Deployment</b></a> · <a href="#gguf-variants"><b>📦 GGUF Variants</b></a> · <a href="https://www.audarai.com/license/audarai-community-license-v1.0/"><b>📜 License</b></a></p> </div>

---

Audar-TTS-V1-Turbo is the balanced production tier of the Audar-TTS family — a ~1.64B-parameter community-licensed speech model that trades a little of Flash's raw speed for higher fidelity and steadier prosody, while staying efficient. It is Arabic-first (including Gulf/Emirati and other dialects), fully bilingual with English, and clones any voice from a 5–15 second reference clip with no per-speaker fine-tuning. This repository ships it as GGUF quantizations for llama.cpp.

Turbo treats speech synthesis as next-token prediction: a language-model backbone predicts discrete audar-codec acoustic tokens, which a lightweight neural codec decodes to 24 kHz audio. There is no phonemizer and no per-language G2P — so Arabic dialects and Arabic⇄English code-switching are handled from data rather than brittle pronunciation rules.

GGUF weights under the AudarAI Community License v1.0 — free for research and limited commercial use by smaller organizations; larger/enterprise use needs an AudarAI Enterprise License. See License.

Highlights

🎧 Voice Gallery

Six ready-to-use voices ship with Audar-TTS, free to use. They are synthetic voices created by interpolating multiple speakers — they do not replicate or resemble any real individual. Each sample is a zero-shot clone: the same reference voice speaks a fresh English and Arabic line. (Captions match the audio word-for-word.)

<table> <thead><tr><th align="left" width="130">Voice</th><th align="left">Reference &amp; samples — Audar-TTS Turbo</th></tr></thead> <tbody> <tr> <td valign="top" width="130"><b>demo_male_1</b><br>_Male<br>_{<i>warm, confident</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_1_source.wav"></audio><br><br><b>English</b> &nbsp; Oh, you have to hear this — [excited] we just closed the biggest deal of the entire year, and honestly, I still can't quite believe it!<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_1_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">لا يمكنني الانتظار لأخبرك — [excited] لقد أنجزنا المشروع أخيراً بعد كلّ هذا التعب، [laughs] وصدّقني، إنه أجمل شعورٍ على الإطلاق!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_1_ar.wav"></audio></td> </tr> <tr> <td valign="top" width="130"><b>demo_male_2</b><br>_Male<br>_{<i>soft, intimate</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_2_source.wav"></audio><br><br><b>English</b> &nbsp; Come a little closer for a second — [whispers] I've been planning something special all week long, [mischievously] and you are going to absolutely love it.<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_2_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">تعال، اقترب قليلاً — [whispers] لقد خطّطتُ لمفاجأةٍ رائعة طوال الأسبوع، [mischievously] وأنا واثقٌ تماماً أنّها ستُدهشك حقاً!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_2_ar.wav"></audio></td> </tr> <tr> <td valign="top" width="130"><b>demo_male_3</b><br>_Male<br>_{<i>bright, curious</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_3_source.wav"></audio><br><br><b>English</b> &nbsp; Wait, really? [curious] You built the whole thing yourself over the weekend? [excited] That is genuinely incredible — tell me everything, right now!<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_3_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">لحظة، حقاً؟ [curious] هل بنيتَ كلّ هذا بنفسك في يومين فقط؟ [excited] هذا مذهلٌ فعلاً — احكِ لي كلّ التفاصيل الآن!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_male_3_ar.wav"></audio></td> </tr> <tr> <td valign="top" width="130"><b>demo_female_1</b><br>_Female<br>_{<i>vibrant, joyful</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_1_source.wav"></audio><br><br><b>English</b> &nbsp; Guess what just arrived in the mail — [excited] the acceptance letter we have been waiting for, [laughs] and I actually screamed out loud!<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_1_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">خمّن ماذا وصل في البريد للتوّ — [excited] رسالة القبول التي انتظرناها طويلاً، [laughs] لقد صرختُ من شدّة الفرح!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_1_ar.wav"></audio></td> </tr> <tr> <td valign="top" width="130"><b>demo_female_2</b><br>_Female<br>_{<i>velvety, playful</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_2_source.wav"></audio><br><br><b>English</b> &nbsp; Okay, lean in for just a moment — [whispers] I found the most perfect little café downtown, [mischievously] and it is going to be our new secret spot.<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_2_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">حسناً، اقتربي قليلاً — [whispers] وجدتُ مقهىً صغيراً رائعاً في وسط المدينة، [mischievously] وسيكون مكاننا السريّ الجديد!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_2_ar.wav"></audio></td> </tr> <tr> <td valign="top" width="130"><b>demo_female_3</b><br>_Female<br>_{<i>airy, dreamy</i>}</td> <td valign="top">_{<b>REFERENCE VOICE</b>}<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_3_source.wav"></audio><br><br><b>English</b> &nbsp; You won't believe the view from up here — [excited] the whole city is glowing at sunset, [laughs] it honestly looks just like a dream!<br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_3_en.wav"></audio><br><br><b>العربية</b> &nbsp; <span dir="rtl">لن تصدّقي هذا المنظر من هنا — [excited] المدينة كلّها تتلألأ عند الغروب، [laughs] وكأنّها لوحةٌ من حلمٍ جميل!</span><br><audio controls preload="none" src="https://huggingface.co/audarai/Audar-TTS-V1-Turbo/resolve/main/samples/demo_female_3_ar.wav"></audio></td> </tr> </tbody> </table>

_{🎛️ <b>Demo generation settings:</b> <code>temperature = 1.0</code> · <code>repetition_penalty = 1.1</code> · <code>top_k = 40</code> · <code>top_p = 0.9</code> — tuned for maximum expressiveness (<code>[laughs]</code>, <code>[excited]</code>, <code>[whispers]</code>…). A low <code>repetition_penalty</code> (≈1.1) is what lets laughter through — a higher value suppresses it. For steadier, more neutral delivery, lower <code>temperature</code> toward <code>0.6–0.7</code>.}

Model summary

<table> <tbody> <tr><td width="190"><b>Model</b></td><td>Audar-TTS-V1-Turbo (GGUF)</td></tr> <tr><td><b>Task</b></td><td>Text-to-speech (autoregressive, neural-codec)</td></tr> <tr><td><b>Backbone</b></td><td>Qwen2.5-1.5B-class decoder-only transformer</td></tr> <tr><td><b>Parameters</b></td><td>1,643,998,208 (1.64B)</td></tr> <tr><td><b>Distribution</b></td><td>GGUF — Q4_K_M / Q5_K_M / Q8_0</td></tr> <tr><td><b>Vocabulary</b></td><td>217,225 (text + 65,536 audar-codec speech tokens + control tokens)</td></tr> <tr><td><b>Context length</b></td><td>32,768 tokens</td></tr> <tr><td><b>Companion codec</b></td><td>audar-codec (a NeuCodec fine-tuned for Arabic) → <b>24 kHz</b> output</td></tr> <tr><td><b>Languages</b></td><td>Arabic (MSA + dialects incl. Gulf/Emirati) and English</td></tr> <tr><td><b>License</b></td><td>AudarAI Community License v1.0</td></tr> </tbody> </table>

The Audar-TTS family

All tiers share one prompt/conditioning protocol, so you can move between them without changing your integration.

GGUF variants

The codec — audar-codec

The backbone emits discrete <|speech_N|> acoustic tokens; a codec turns those into a 24 kHz waveform. These tokens are decoded by audar-codec — Audar's fine-tuned NeuCodec, adapted for Arabic on extensive data.

🙏 Credit & thanks to Neuphonic for open-sourcing NeuCodec. audar-codec builds on their work, and the tokens remain NeuCodec-compatible — so you can decode with NeuCodec directly (as shown below), which makes this release fully open and reproducible.

Local deployment (GGUF)

# pip install llama-cpp-python neucodec soundfile torch huggingface_hub
import re, torch, soundfile as sf
from huggingface_hub import hf_hub_download
from llama_cpp import Llama
from neucodec import NeuCodec   # base NeuCodec (public); audar-codec is the Arabic-tuned companion

# 1) Backbone (GGUF) — CPU by default; set n_gpu_layers=-1 to offload to GPU
gguf = hf_hub_download("audarai/Audar-TTS-V1-Turbo", "Audar-TTS-V1-Turbo-Q4_K_M.gguf")
llm  = Llama(model_path=gguf, n_ctx=4096, n_gpu_layers=0, verbose=False)

# 2) Codec — encodes the reference clip and decodes the output
codec = NeuCodec.from_pretrained("neuphonic/neucodec").eval()

# 3) Zero-shot reference: a 5-15 s clip (16 kHz mono) + its transcript
ref_codes = codec.encode_code("reference.wav").squeeze().tolist()
ref_text  = "transcript of the reference clip"
ref = "".join(f"<|speech_{c}|>" for c in ref_codes)

target = "مرحبا! [whispers] أهلاً وسهلاً بك."
prompt = ("user: Convert the text to speech:"
          f"<|REF_TEXT_START|>{ref_text}<|REF_TEXT_END|>"
          f"<|REF_SPEECH_START|>{ref}<|REF_SPEECH_END|>"
          f"<|TARGET_TEXT_START|>{target}<|TARGET_TEXT_END|>"
          "\nassistant:<|TARGET_CODES_START|>")

# 4) Generate speech tokens; stop at <|TARGET_CODES_END|>
tce  = llm.tokenize(b"<|TARGET_CODES_END|>", add_bos=False, special=True)[0]
toks = llm.tokenize(prompt.encode("utf-8"), add_bos=False, special=True)
ids  = []
for tid in llm.generate(toks, temp=1.0, top_k=40, top_p=0.9, repeat_penalty=1.1):
    if tid == tce or len(ids) >= 2048: break
    ids.append(tid)
text = "".join(llm.detokenize([t], special=True).decode("utf-8", "ignore") for t in ids)

# 5) Decode to 24 kHz audio
codes = [int(x) for x in re.findall(r"<\|speech_(\d+)\|>", text)]
wav = codec.decode_code(torch.tensor(codes)[None, None, :]).cpu().numpy()[0, 0, :]
sf.write("out.wav", wav, 24000)

Prefer a managed endpoint? The same model is available via the Audar API/SDK (client.tts, model id audar-tts-v1-turbo).

Expression tags

Insert tags inline in the target text to shape delivery:

[laughs] · [curious] · [excited] · [sighs] · [exhales] · [mischievously] · [whispers] · [sarcastic]

Tags work in both Arabic and English. Use them sparingly for the most natural results.

Intended use & limitations

Intended use. Voice assistants and agents, narration, accessibility, IVR/telephony, and any application that synthesizes Arabic and English speech — in the cloud, on-premises, or offline. Zero-shot cloning is intended for consented voices only.

Limitations.

• Chunk very long inputs at sentence boundaries.
• Multi-word or stacked expression tags are more fragile than single tags.
• As with all neural TTS, rare names, numbers, and code-switch boundaries can be mispronounced.

Evaluation

Audar-TTS is evaluated on an internal cross-provider Arabic TTS benchmark covering intelligibility (resynthesis WER/CER), predicted naturalness (UTMOSv2, SQUIM), speaker similarity, and expression fidelity. We report results with their caveats: predicted-MOS metrics are not calibrated for Arabic, similarity/expression proxies are English-biased, and a formal human MOS/CMOS study is planned, not yet completed. Qualitatively, Audar-TTS is designed to be competitive with leading systems on Arabic intelligibility and expression control, with particular strength on Gulf-dialect speech.

📊 An audited benchmark table will accompany the forthcoming Audar-TTS technical report.

📜 License

Released under the AudarAI Community License, Version 1.0 — full text: AudarAI Community License v1.0 · repo copy. This is source-available and open-weight, but not unrestricted. In brief (the license text controls):

• ✅ Research use — free for research, evaluation, benchmarking, education, and prototyping.
• ✅ Limited commercial use for a Community Entity — an organization within all of these thresholds: < 50 staff, < $2M annual revenue, < $5M lifetime funding, < 100k monthly active users, and < $250k revenue attributable to the model (§3).
• ✅ Modify, fine-tune, quantize, prune, merge, adapt for permitted uses (§2.4).
• ⬆️ Enterprise use needs an AudarAI Enterprise License (contact sales) — required if you exceed any threshold, or for enterprise/production-at-scale deployment, model-as-a-service or APIs, High-Risk uses, or training a competing foundation model (§4, §5).
• 📋 Conditions: don't redistribute the original weights except by linking to the official repo (§8.2); pass on the license + notices with any modified redistribution; no trademark rights; and follow the Acceptable Use Restrictions (Schedule A).

The backbone is adapted from Qwen2.5 (Apache-2.0); the audar-codec decoder is distributed separately and builds on NeuCodec (see §15, Third-Party Components).

Note: the smaller Audar-TTS-V1-Flash tier is released under the more permissive AudarAI Open License v1.0 (unrestricted commercial use).

Responsible use

Clone voices only with the speaker's explicit consent. Do not use Audar-TTS to deceive, defraud, or impersonate real people or organizations, and comply with applicable law (§9).

Citation

@misc{audar-tts-turbo-2026,
  title  = {Audar-TTS: Arabic-First Expressive Speech Synthesis},
  author = {Audar AI Research Team},
  year   = {2026},
  note   = {Audar-TTS-V1-Turbo},
  url    = {https://huggingface.co/audarai/Audar-TTS-V1-Turbo}
}

---

About AudarAI

<div align="center">

Leading Arabic-First Multilingual Audio Intelligence

AudarAI starts with Arabic — and expands to the world.

</div>

We are building advanced multilingual audio intelligence that helps individuals, enterprises, and governments communicate across languages, cultures, and borders. By combining Arabic-first speech technology with global multilingual AI, AudarAI transforms voice into understanding, interaction, and connection.

Our work spans speech recognition, speech understanding, voice-enabled digital assistants, human-computer interaction, and intelligent audio systems designed for real-world impact. From empowering people to access technology in their native language to helping organizations communicate globally, AudarAI is shaping a future where every voice can be heard, understood, and connected.

Arabic-first. Multilingual by design. Human-centered at heart.

<div align="center">

🌐 www.audarai.com · 🤗 Hugging Face · GitHub · [email protected]

© 2026 AUDARAI PTE. LTD. · Licensed under the AudarAI Community License v1.0

</div>