Skip to content
Audio

Chatterbox-Persian-TTS

Persian TTS without G2P

2 min readupdated

This model is a full fine-tune of the Chatterbox Multilingual architecture, trained on a large 417-hour Persian speech dataset. It takes raw Persian text directly as input, skipping phonetic segmentation and G2P (grapheme-to-phoneme) systems altogether. Built for developers and researchers, it delivers high-quality Persian text-to-speech alongside short-sample voice cloning.

Key features

  • Full fine-tune of the base model on 417 hours of Persian speech (roughly 109,000 audio clips).
  • Direct processing of raw Persian text without external G2P modules.
  • Built-in voice cloning from a clean 5–10 second audio reference.
  • Stable generation defaults using lower temperature to cut down on artifacts.

What you can do

  • Generate natural Persian speech from text using the ready-to-run Google Colab notebook.
  • Build a voice assistant or text-to-speech bot for Persian.
  • Clone a speaker's vocal timbre from a short reference recording.

Example

import torch
from chatterbox.mtl_tts import ChatterboxMultilingualTTS, punc_norm
from safetensors.torch import load_file

model = ChatterboxMultilingualTTS.from_pretrained(device="cuda")
sd = load_file("t3_persian_v1.safetensors")
model.t3.load_state_dict(
    {k[3:] if k.startswith("t3.") else k: v for k, v in sd.items()},
    strict=False,
)
model.t3.to("cuda").eval()

wav = model.generate(
    punc_norm("سلام، امروز هوا خوب است."),
    language_id=None,
    audio_prompt_path="ref.wav",
    temperature=0.6,
    repetition_penalty=1.5,
)
Read moreDetails
Benchmarks (CER): 0.01–0.09 on natural text, and 0.038–0.062 on harder tests containing punctuation and numbers.
Output sample rate: 16 kHz.
License: MIT.
Limitations: Occasional hallucinations and repetitions can occur, requiring a re-run.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 2.0 GB · ready files2.7 GB
  • FP8 / INT8recommendedweights 1.0 GB1.6 GB
  • 4-bitweights 0.6 GB1.1 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # amirzandi/Chatterbox-Persian-TTS
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download amirzandi/Chatterbox-Persian-TTS --local-dir ~/models/Chatterbox-Persian-TTS
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Chatterbox-Persian-TTS
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of chatterbox (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 5All on Hugging Face
Merges 1All on Hugging Face

About the model

amirzandi/Chatterbox-Persian-TTS
Author
amirzandi
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments