Skip to content
Audio

AzurLane GPT-SoVITS v2ProPlus: 902 Character Voice Pack

2 min read

This set of weights for GPT-SoVITS (a voice cloning and speech synthesis system) is trained to voice 902 ships and TB assistants from Azur Lane in Japanese and Chinese. A total of 933 voice presets are available, including alternative versions and different localization options. The model is based on the official GPT-SoVITS v2ProPlus weights and allows switching characters by simply changing the audio preset while using shared base weights. This release is useful for game content creators, video authors, and fans of media projects.

Key features

  • Covers 902 characters and 933 ready-made voice presets (912 Japanese and 21 Chinese).
  • Built on GPT-SoVITS v2ProPlus with improved intonation transfer quality.
  • Separate checkpoints: the GPT module is optimized separately for Japanese (epoch 12) and Chinese (epoch 14) speech, while a shared SoVITS is used (epoch 8).
  • Full compatibility with the Neiroha-GPT-SoVITS pipeline and the standard WebUI.

What you can do

  • Voice Azur Lane character lines in Japanese or Chinese for fan videos and mods.
  • Run a local speech synthesis server via the Neiroha-GPT-SoVITS Windows build.
  • Configure voice switching via a configuration .toml file using character IDs (for example, for Nagato or TB).

Example

$gptEpoch = 12  # Для китайской озвучки укажите 14
hf download Moear/AzurLane-GPT-SoVITS-v2ProPlus `
  "all-characters/s1/azur-lane-all-v2proplus-20261119-e$gptEpoch.ckpt" `
  all-characters/s2/azur-lane-all-v2proplus-20261119_e8_s12712.pth `
  --local-dir .\models\voices\azurlane
Read moreDetails
• Dataset: 50,759 entries (about 138.3 hours of in-game audio).
• Quality: in ERes2NetV2 tests, cosine similarity reaches 0.8442 for Japanese speech (12+8) and 0.8230 for Chinese (14+8).
• Performance: on an RTX 6000D, the synthesis time for short phrases is about 0.25 of the audio duration.
• Limitations: requires downloading configuration presets and editing paths in the .toml file.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.2 GB · ready files0.7 GB
  • FP8 / INT8recommendedweights 0.1 GB0.6 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.

    # liujianzuo/AzurLane-GPT-SoVITS-v2ProPlus
    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 liujianzuo/AzurLane-GPT-SoVITS-v2ProPlus --local-dir ~/models/AzurLane-GPT-SoVITS-v2ProPlus
    # 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 GPT-SoVITS (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 5All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 2All on Hugging Face

About the model

liujianzuo/AzurLane-GPT-SoVITS-v2ProPlus
Author
liujianzuo
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

Audio

pingo-ptbr

pingo-ptbr