Skip to content
Other

PC-NSF-HiFiGAN 2025.02

new official vocoder from OpenVPI

2 min readupdated

This is the official PC-NSF-HiFiGAN vocoder (a tool that synthesizes an audio waveform from a spectrogram) from the 2025.02 release by the OpenVPI team and the DiffSinger community. It runs at a 44.1 kHz sample rate with a hop size of 512 and 128 mel bins (frequency bands used for audio analysis), and ships in its original, unmodified ONNX format. Built for speech synthesizers and DiffSinger-based music projects, it delivers clean, detailed speech and vocals.

Key features

  • Unmodified official 2025.02 release from the OpenVPI team and the DiffSinger community.
  • Delivers high-fidelity audio at a 44.1 kHz sample rate.
  • Ships in ONNX format alongside a vocoder.yaml parameter configuration file.
  • Licensed under Attribution-NonCommercial-ShareAlike 4.0 International for non-commercial components.

What you can do

  • Integrate the vocoder into speech or singing synthesis software built on DiffSinger.
  • Drop the model into audio generation pipelines for clean 44.1 kHz output.
  • Tune hyperparameters via the included vocoder.yaml config file to match specific synthesis tasks.

Example

The model ships as a ready-to-use pc_nsf_hifigan_44.1k_hop512_128bin_2025.02.onnx file paired with a vocoder.yaml config, ready to load into compatible speech synthesis projects that support OpenVPI vocoders.

Read moreDetails
Sample rate: 44.1 kHz
Parameters: hop 512, 128 mel bins
Format: ONNX
License: CC BY-NC-SA 4.0 (non-commercial component, though audio output is unrestricted)
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.1 GB · ready files0.6 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.

    # AEmotionStudio/openvpi-pc-nsf-hifigan-2025-02
    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 AEmotionStudio/openvpi-pc-nsf-hifigan-2025-02 --local-dir ~/models/openvpi-pc-nsf-hifigan-2025-02
    # 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 openvpi-pc-nsf-hifigan-2025-02 (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

AEmotionStudio/openvpi-pc-nsf-hifigan-2025-02
Author
AEmotionStudio
Category
Other
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