Skip to content
Audio

Fluent Books: English voices r6

восемь компактных TTS-голосов для чтения книг

2 min readupdated

Это восемь небольших text-to-speech моделей в формате ONNX для приложения Fluent Books: чтение книги вслух на процессоре телефона. Четыре американских и четыре британских голоса, каждый файл — отдельный голос с тембром, зашитым в веса. Это student-модели, дистиллированные из английских голосов Kokoro-82M; размер каждого файла — 8.57 MB.

Ключевые особенности

  • Восемь голосов: af_heart, af_kore, am_michael, am_fenrir, bf_emma, bf_isabella, bm_george, bm_fable.
  • Каждый ONNX-файл — один голос; тембр зашит в веса, размер — 8.57 MB.
  • На вход подаются Kokoro phoneme ids и speed, на выходе — waveform 24 кГц и durations по токенам.
  • Один student на голос, дистилляция из английских голосов Kokoro-82M.
  • Голоса — новые версии голосов приложения; доступны рядом со старыми под id <voice>.r6.
  • Текст нужно нормализовать заранее: числа, сокращения, даты.

Пример

Для голоса af_heart.r6 на вход подаются Kokoro phoneme ids в формате [0] + ids + [0] (int64, [1,N]) и speed=[1.0] (float32, [1]). На выходе — waveform [1,S] с частотой 24 кГц и durations [1,T], где S = 600 * sum(durations).

ПодробнееРазмер одного файла — 8.57 MB, запуск на CPU телефона.
UTMOS 3.98–4.43, WER 2.6–3.4%, voice similarity 0.956–0.973 на 60 отложенных предложениях.
Детектор артефактов отмечает особенности у большинства голосов; у bm_george — шесть.
Происхождение данных голосов Kokoro-82M не подтверждено авторами репозитория; веса заявлены как Apache-2.0.
Текст нужно нормализовать заранее; тренировка — ноябрь 2026.
Discuss Telegram 0

Will it run?

Runs in INT8

Fits without quantisation.

  • INT8recommendedweights 0.0 GB · ready files0.5 GB
  • 4-bitweights 0.0 GB0.5 GB

Tip: go 8-bit.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: INT8 — 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.

    # lazarenkod/fluent-books-voices-en
    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 lazarenkod/fluent-books-voices-en --local-dir ~/models/fluent-books-voices-en
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to fluent-books-voices-en
  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 fluent-books-voices-en (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

lazarenkod/fluent-books-voices-en
Author
lazarenkod
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