Skip to content
Audio

Audio8-TTS-Realistic-Air

компактный TTS на 23 языка с реалистичной речью и клонированием голоса

2 min read

Audio8-TTS-Realistic-Air — обновление Edge0/Audio8-TTS-Preview-0.6b от TechnoBaptist: TTS-модель на 0.6B параметров (601 159 424 без кодека) с zero-shot клонированием голоса. Сохраняет компактную DualAR-архитектуру, но обещает заметно более реалистичную речь; поддерживает 23 языка, включая русский.

Ключевые особенности

  • Реализм: более естественная просодия и выразительность по сравнению с Preview.
  • 23 языка: 11 из Preview плюс 12 новых, включая русский, арабский, хинди и португальский.
  • Zero-shot клонирование по референсной записи и точному транскрипту; есть генерация без референса.
  • DualAR: slow AR предсказывает semantic token на кадр, fast AR — codebooks кодека; 10 codebooks по 4096 записей.
  • SGLang Omni: paged attention, dynamic batching и OpenAI-совместимый /v1/audio/speech.
  • В комплекте 44.1 kHz кодек, tokenizer, processor и кастомный код Transformers.

Пример

import torch
from transformers import AutoModel, AutoProcessor
model_id = "Edge0/Audio8-TTS-Realistic-Air"
processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
model = AutoModel.from_pretrained(model_id, trust_remote_code=True, dtype=torch.bfloat16).eval().cuda()
inputs = processor(text=["Hello from Audio8."], reference_audio=["ref.wav"], reference_text=["Exact transcript."], return_tensors="pt")
inputs = {k: v.to("cuda") for k, v in inputs.items()}
output = model.generate(**inputs, max_new_tokens=1024, do_sample=True)
ПодробнееSeed-TTS EN/ZH: error 1.25%, SIM 0.704; CV3 (9 языков): 4.86%, SIM 0.661; MiniMax (24 языка): 5.89%, SIM 0.754.
Рекомендуются Python 3.10+ и CUDA-GPU; smoke-тест шёл в BF16 на NVIDIA H20. VRAM не указан.
Качество зависит от языка: на CV3 Thai WER 48.574%, Arabic 21.098%; стриминг и throughput не бенчмаркались.
Discuss Telegram 1 26

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.1 GB · ready files1.7 GB
  • FP8 / INT8recommendedweights 0.6 GB1.1 GB
  • 4-bitweights 0.3 GB0.8 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.

    # TechnoBaptist/Audio8-TTS-Realistic-Air
    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 TechnoBaptist/Audio8-TTS-Realistic-Air --local-dir ~/models/Audio8-TTS-Realistic-Air
    # The model is in ~/models. How to run it is in the model card (link above).
  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 Audio8-TTS-Preview-0.6b (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 9All on Hugging Face
Fine-tunes 5All on Hugging Face

About the model

TechnoBaptist/Audio8-TTS-Realistic-Air
Author
TechnoBaptist
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

261

Ask AI about this model

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

Comments