Skip to content
Audio

DAW Vocal Denoiser v1d

a lightweight model for real-time vocal cleaning

2 min readupdated

DAW Vocal Denoiser v1d is a compact model (1.25 million parameters) based on a GRU (recurrent neural network) designed to remove noise from vocal tracks and speech. Version v1d adds 5 ms of lookahead, achieving a total algorithmic latency of 15 ms. This architecture delivers quality comparable to models twice its size while requiring fewer CPU resources. The tool is useful for musicians and sound engineers for real-time monitoring and recording cleanup.

Key features

  • Low latency: an algorithmic latency of only 15 ms, suitable for real-time monitoring.
  • Efficiency: quality on par with models twice as heavy at half the CPU requirements.
  • Compactness: the ONNX file size is just 5 MB.
  • Sound quality: high cleaning accuracy without damaging clean audio sections.

What you can do

  • Vocal cleanup: integrate the model into a DAW (digital audio workstation) to suppress room noise during recording.
  • Docker usage: run a container for batch file processing or set up a local HTTP server.
  • Python integration: use ONNX Runtime for streaming or offline audio processing.

Example

An example of using the model for streaming audio processing in Python:

# pip install onnxruntime numpy
from stream import StreamEngine          # ожидает daw_denoiser.onnx рядом
eng = StreamEngine()
y_block = eng.process(x_block)           # float32[240] на 48 кГц -> очищенные семплы

from stream import process_file          # офлайн-обработка массива
y = process_file(x)
Read moreDetails
• SI-SDR: 23.43 dB, SDR: 24.03 dB, PESQ-WB: 3.11
• Requirements: 0.23 ms CPU per 5 ms block, ONNX format (5.0 MB)
• Limitations: mono only, 48 kHz; removes noise, not reverberation
• Training: 108,779 steps on RTX 4060
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 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.

    # Stavion-Colquitt/DAW_Vocal_Denoiser_v1d
    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 Stavion-Colquitt/DAW_Vocal_Denoiser_v1d --local-dir ~/models/DAW_Vocal_Denoiser_v1d
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to DAW_Vocal_Denoiser_v1d
  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 DAW_Vocal_Denoiser_v1d (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

Stavion-Colquitt/DAW_Vocal_Denoiser_v1d
Author
Stavion-Colquitt
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