C1Tech/VibeVoice_Persian

VibeVoice-Persian is an advanced Persian Text-To-Speech (TTS) model fine-tuned on top of microsoft/VibeVoice-1.5B. Designed specifically to bring expressive, highly natural, and context-aware speech synthesis to the Persian language, this model leverages a high-quality custom dataset to achieve rich prosody and fluid conversational delivery.

Key Features

• Native Persian Prosody: Fine-tuned to capture natural Persian accentuation, cadence, and intonation patterns.
• Zero-Shot Voice Cloning: Prompt the model with short audio references (.wav) to replicate tone and speaker characteristics.
• Bilingual Support: Robust performance on Persian text, with smooth handling of embedded English words and mixed-language phrases.

Samples

Usage

git clone https://github.com/vibevoice-community/VibeVoice.git
cd VibeVoice && pip install -e .
pip install torch torchaudio huggingface_hub

import torch, torchaudio
from vibevoice.modular.modeling_vibevoice_inference import VibeVoiceForConditionalGenerationInference
from vibevoice.processor.vibevoice_processor import VibeVoiceProcessor

model_id = "C1Tech/VibeVoice_Persian"
device = "cuda" if torch.cuda.is_available() else "cpu"

model = VibeVoiceForConditionalGenerationInference.from_pretrained(
    model_id, torch_dtype=torch.bfloat16, device_map=device
)
model.eval()
model.set_ddpm_inference_steps(num_steps=20)
processor = VibeVoiceProcessor.from_pretrained(model_id)
text = "سلام آرش، حالت چطوره؟"
inputs = processor(
    text=["Speaker 1: "+text],
    voice_samples=[["ref.wav"]],
    speaker_names=[["Ref"]],
    return_tensors="pt", return_attention_mask=True,
)
inputs = {k: v.to(device) if torch.is_tensor(v) else v for k, v in inputs.items()}
with torch.no_grad():
    out = model.generate(**inputs, cfg_scale=1.5, tokenizer=processor.tokenizer,
                          generation_config={"do_sample": False})

audio = out.speech_outputs[0].float().cpu().squeeze()
torchaudio.save("output.wav", audio.unsqueeze(0), 24000)

Responsible Usage

سلب مسئولیت (Disclaimer)

این مدل صرفاً برای اهداف پژوهشی و آموزشی در حوزه‌ی تولید گفتار منتشر شده است. شرکت سیوان‌تک هیچ‌گونه مسئولیتی را در قبال نحوه‌ی استفاده، سوءاستفاده، بهره‌برداری غیراخلاقی یا غیرمجاز از این مدل توسط کاربران یا اشخاص ثالث نمی‌پذیرد. مسئولیت کامل استفاده از خروجی‌های تولیدشده توسط این مدل، از جمله رعایت قوانین و مقررات جاری، حقوق مالکیت معنوی، حریم خصوصی افراد و اصول اخلاقی، صرفاً بر عهده‌ی کاربر نهایی است. هرگونه استفاده از این مدل برای جعل هویت صوتی افراد بدون رضایت صریح و مستند آن‌ها، تولید اخبار یا محتوای گمراه‌کننده، فریب یا کلاه‌برداری، و یا هر فعالیت غیرقانونی دیگر، به‌شدت ممنوع بوده و خارج از حدود مجوز استفاده از این مدل است. شرکت سیوان‌تک در برابر هرگونه خسارت مستقیم یا غیرمستقیم ناشی از استفاده از این مدل، تحت هیچ شرایطی مسئولیتی نخواهد داشت.

Direct intended uses

The Persian VibeVoice model is limited to research purpose use exploring highly realistic audio dialogue generation.

Out-of-scope uses

Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by MIT License. Use to generate any text transcript. Furthermore, this release is not intended or licensed for any of the following scenarios:

• Voice impersonation without explicit, recorded consent – cloning a real individual’s voice for satire, advertising, ransom, social‑engineering, or authentication bypass.
• Disinformation or impersonation – creating audio presented as genuine recordings of real people or events.
• Real‑time or low‑latency voice conversion – telephone or video‑conference “live deep‑fake” applications.
• Unsupported language – the model is trained only on Persian and English data; outputs in other languages are unsupported and may be unintelligible or offensive.
• Generation of background ambience, Foley, or music – VibeVoice is speech‑only and will not produce coherent non‑speech audio.

Risks and limitations

While efforts have been made to optimize it through various techniques, it may still produce outputs that are unexpected, biased, or inaccurate. VibeVoice inherits any biases, errors, or omissions produced by its base model. Potential for Deepfakes and Disinformation: High-quality synthetic speech can be misused to create convincing fake audio content for impersonation, fraud, or spreading disinformation. Users must ensure transcripts are reliable, check content accuracy, and avoid using generated content in misleading ways. Users are expected to use the generated content and to deploy the models in a lawful manner, in full compliance with all applicable laws and regulations in the relevant jurisdictions. It is best practice to disclose the use of AI when sharing AI-generated content. Persian and English only: Transcripts in language other than Persian or English may result in unexpected audio outputs. Non-Speech Audio: The model focuses solely on speech synthesis and does not handle background noise, music, or other sound effects. Overlapping Speech: The current model does not explicitly model or generate overlapping speech segments in conversations.

---

We specialize in cutting-edge Voice & Audio Intelligence—from state-of-the-art Speech-to-Text (STT) and natural Text-to-Speech (TTS) to Voice Verification, Audio Intelligence, and domain-adapted LLMs.

Need higher accuracy, lower latency, or custom-trained voice models for your enterprise?

📬 Contact Sales: [email protected]

🔗 Explore Dashboard: https://ai.c1tech.group/dashboard