Skip to content
Audio

Resemble Enhance LiteRT

очистка речи локально и на Android

1 min readupdated

Экспериментальная LiteRT-конверсия модели Resemble Enhance от SayedShaun для локального шумоподавления и улучшения речи. Включает пять LiteRT-моделей: веса хранятся в float16, входы, выходы и вычисления на CPU — float32. Общий размер — около 712 MB.

Ключевые особенности

  • Локальная очистка и улучшение речи без облака.
  • Python-инференс через ai-edge-litert, NumPy и SciPy; PyTorch не нужен.
  • Есть Kotlin-API для Android: файлы из models/ копируются на устройство, запуск — в worker thread.
  • На вход можно подать mono или массив NumPy (frames, channels); выход — mono float32 44.1 kHz.
  • Обработка идёт перекрывающимися окнами примерно по 1 секунде.

Пример

import soundfile as sf
from inference import ResembleEnhancer

model = ResembleEnhancer('SayedShaun/resemble-enhance-litert')
audio, sample_rate = model.enhance('input.wav')
sf.write('enhanced.wav', audio, sample_rate)
ПодробнееВсего пять LiteRT-моделей, общий размер ~712 MB; веса float16, вычисления CPU float32.
Python-инференс тестировали на Linux x86-64 с Python 3.12; Android-пайплайн ещё не запускали на устройстве или эмуляторе.
В сравнении с PyTorch на шумовом сэмпле 49 из 44 100 значений превысили atol=0.002, rtol=0.002; максимальная разница — 0.00378.
Чанки по 1 секунде могут давать артефакты; это экспериментальная конверсия, а не официальный релиз Resemble AI.
Discuss Telegram 0 102

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

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.

    # SayedShaun/resemble-enhance-litert
    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 SayedShaun/resemble-enhance-litert --local-dir ~/models/resemble-enhance-litert
    # 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 resemble-enhance-litert (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

SayedShaun/resemble-enhance-litert
Author
SayedShaun
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

1020

Ask AI about this model

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

Comments