Tsukuyomichan OmniVoice Full Finetune

Japanese / ๆ—ฅๆœฌ่ชž

  • ๆ—ฅๆœฌ่ชž็‰ˆREADME: README_ja.md
  • ใ“ใฎๅ…ฌ้–‹็‰ฉใฏ k2-fsa/OmniVoice ใ‚’ใคใใ‚ˆใฟใกใ‚ƒใ‚“่ฉฑ่€…ๅ‘ใ‘ใซใƒ•ใƒซใƒ•ใ‚กใ‚คใƒณใƒใƒฅใƒผใƒ‹ใƒณใ‚ฐใ—ใŸ checkpoint ใงใ™
  • ็พๆ™‚็‚นใงใ€ๅ†…้ƒจๆฏ”่ผƒใงใฏใ“ใฎ full finetune ็‰ˆใŒๆœ€ๆŽจๅฅจใงใ™

English

This repository provides the full finetuned OmniVoice checkpoint for a Tsukuyomichan-like Japanese TTS voice.

At the current stage, this full finetune variant is the recommended release over the LoRA variants.

Files

  • model.safetensors: finetuned model weights
  • config.json: model configuration
  • tokenizer.json: tokenizer file
  • tokenizer_config.json: tokenizer configuration
  • chat_template.jinja: tokenizer chat template
  • train_config.json: training configuration used for this release
  • eval_summary.json: OpenVoice-based speaker similarity summary

Installation

The easiest way to use this checkpoint is with omnivoice-kit.

git clone https://github.com/kizuna-intelligence/omnivoice-kit
cd omnivoice-kit
git submodule update --init --recursive
python -m venv .venv
source .venv/bin/activate
pip install -e third_party/OmniVoice
pip install -e .

How This Model Was Trained

This release was trained from a processed Tsukuyomichan VOICEACTRESS100_###/*.flac + JSON corpus using full finetuning.

Core settings:

  • initialization checkpoint: k2-fsa/OmniVoice
  • schedule:
    • 0 -> 300 step: learning_rate=2e-5
    • 300 -> 450 step: learning_rate=1e-5
    • 450 -> 500 step: learning_rate=3e-6
  • batch_tokens=256
  • gradient_accumulation_steps=4
  • mixed_precision=bf16
  • prompt_ratio_range=[0.0, 0.0]
  • drop_cond_ratio=0.0
  • mask_ratio_range=[0.2, 1.0]

The exact config is included as train_config.json.

Inference With This Checkpoint

omnivoice-kit generate \
  --base-model k2-fsa/OmniVoice \
  --checkpoint-dir /path/to/this-checkpoint \
  --input-jsonl examples/japanese_prompts.jsonl \
  --output-dir artifacts/generate_ja \
  --language ja \
  --num-step 16

Recommended inference setting:

  • num_step=16

Tsukuyomichan TTS Usage

For Tsukuyomichan-like Japanese TTS, use this checkpoint as the default recommendation.

Internal comparison summary:

  • no-ref gen_to_train_centroid_mean = 0.9022
  • with-ref gen_to_train_centroid_mean = 0.8971

Compared to the base OmniVoice model:

  • base no-ref: 0.4059 -> 0.9022
  • base with-ref: 0.8848 -> 0.8971

Audio Samples

Japanese

Text: ใ‚„ใฃใŸโ€ฆโ€ฆใ‚„ใฃใจ็ต‚ใ‚ใฃใŸใ‚“ใ ใญใ€‚ใกใ‚ƒใ‚“ใจใ“ใ“ใพใงๆฅใ‚‰ใ‚Œใฆใ€ๆœฌๅฝ“ใซใ‚ˆใ‹ใฃใŸใ€‚

Direct file: samples/emotional/wav/05_emo_05_relief.wav

English

Text: You being here makes me feel like everything will be okay.

Direct file: samples/english/no_ref/wav/02_eng_002.wav

Hard Japanese

Text: ๆฑไบฌ็‰น่จฑ่จฑๅฏๅฑ€่จฑๅฏๅฑ€้•ทใ€ไปŠๆ—ฅใ‚‚็‰น่จฑ่จฑๅฏๅฑ€่จฑๅฏๅฑ€้•ทใ€‚

Direct file: samples/hard/wav/02_hard_02_tokkyo.wav

ใ‚ฏใƒฌใ‚ธใƒƒใƒˆใจๆณจๆ„

ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ณใƒผใƒ‘ใ‚น ใ‚ฏใƒฌใ‚ธใƒƒใƒˆ

ๆœฌใƒขใƒ‡ใƒซใฎๅญฆ็ฟ’ใซใฏใ€ใƒ•ใƒชใƒผ็ด ๆใ‚ญใƒฃใƒฉใ‚ฏใ‚ฟใƒผใ€Œใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ€๏ผˆยฉ Rei Yumesaki๏ผ‰ใŒ็„กๆ–™ๅ…ฌ้–‹ใ—ใฆใ„ใ‚‹้Ÿณๅฃฐใƒ‡ใƒผใ‚ฟใ‚’ไฝฟ็”จใ—ใฆใ„ใพใ™ใ€‚

ๆœฌใƒขใƒ‡ใƒซใ‚„ๆœฌใƒขใƒ‡ใƒซใ‚’็”จใ„ใŸใƒ‡ใƒขใƒปใ‚ขใƒ—ใƒชใ‚ฑใƒผใ‚ทใƒงใƒณใ‚’ๅ…ฌ้–‹ใ™ใ‚‹้š›ใฏใ€ไปฅไธ‹ใฎใ‚ฏใƒฌใ‚ธใƒƒใƒˆใ‚’็œ็•ฅใ›ใšๆŽฒ่ผ‰ใ—ใฆใใ ใ•ใ„ใ€‚

ๆœฌใ‚ฝใƒ•ใƒˆใ‚ฆใ‚งใ‚ขใฎ้Ÿณๅฃฐๅˆๆˆใซใฏใ€ใƒ•ใƒชใƒผ็ด ๆใ‚ญใƒฃใƒฉใ‚ฏใ‚ฟใƒผใ€Œใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ€๏ผˆยฉ Rei Yumesaki๏ผ‰ใŒ็„กๆ–™ๅ…ฌ้–‹ใ—ใฆใ„ใ‚‹้Ÿณๅฃฐใƒ‡ใƒผใ‚ฟใ‚’ไฝฟ็”จใ—ใฆใ„ใพใ™ใ€‚

โ– ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ณใƒผใƒ‘ใ‚น๏ผˆCV.ๅคขๅ‰้ปŽ๏ผ‰
https://tyc.rei-yumesaki.net/material/corpus/

ใƒฆใƒผใ‚ถใƒผใธใฎ็Ÿญใ„ๆกˆๅ†…ไพ‹:

้Ÿณๅฃฐๅˆๆˆใซใฏใ€Œใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ณใƒผใƒ‘ใ‚น๏ผˆCV: ๅคขๅ‰้ปŽ๏ผ‰ใ€ใ‚’ไฝฟ็”จใ—ใฆใ„ใพใ™ใ€‚

ๅ‡บๅŠ›้Ÿณๅฃฐใฎๅˆฉ็”จๅˆถ้™

ๆœฌใƒขใƒ‡ใƒซใ‹ใ‚‰ๅ‡บๅŠ›ใ—ใŸ้Ÿณๅฃฐใฏใ€ๆฌกใฎ็›ฎ็š„ใงใฏใ”ๅˆฉ็”จใ„ใŸใ ใ‘ใพใ›ใ‚“ใ€‚

  • ไบบใ‚’ๆ‰นๅˆคใƒปๆ”ปๆ’ƒใ™ใ‚‹ใ“ใจใ€‚๏ผˆใ€Œๆ‰นๅˆคใƒปๆ”ปๆ’ƒใ€ใฎๅฎš็พฉใฏใ€ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ญใƒฃใƒฉใ‚ฏใ‚ฟใƒผใƒฉใ‚คใ‚ปใƒณใ‚น ใซๆบ–ใ˜ใพใ™๏ผ‰
  • ็‰นๅฎšใฎๆ”ฟๆฒป็š„็ซ‹ๅ ดใƒปๅฎ—ๆ•™ใƒปๆ€ๆƒณใธใฎ่ณ›ๅŒใพใŸใฏๅๅฏพใ‚’ๅ‘ผใณใ‹ใ‘ใ‚‹ใ“ใจใ€‚
  • ๅˆบๆฟ€ใฎๅผทใ„่กจ็พใ‚’ใ‚พใƒผใƒ‹ใƒณใ‚ฐใชใ—ใงๅ…ฌ้–‹ใ™ใ‚‹ใ“ใจใ€‚
  • ไป–่€…ใซๅฏพใ—ใฆไบŒๆฌกๅˆฉ็”จ๏ผˆ็ด ๆใจใ—ใฆใฎๅˆฉ็”จ๏ผ‰ใ‚’่จฑๅฏใ™ใ‚‹ๅฝขใงๅ…ฌ้–‹ใ™ใ‚‹ใ“ใจใ€‚

โ€ป้‘‘่ณž็”จใฎไฝœๅ“ใจใ—ใฆ้…ๅธƒใƒป่ฒฉๅฃฒใ—ใฆใ„ใŸใ ใใ“ใจใฏๅ•้กŒใ”ใ–ใ„ใพใ›ใ‚“ใ€‚

ๆ”นๅค‰ใƒปๅ†้…ๅธƒใซใคใ„ใฆ

ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใฎใƒขใƒ‡ใƒซใใฎใ‚‚ใฎใ‚’็ด ๆใจใ—ใฆไฝฟ็”จใ™ใ‚‹ๅ ดๅˆ๏ผˆๆ”นๅค‰ใ€ใƒ•ใ‚กใ‚คใƒณใƒใƒฅใƒผใƒ‹ใƒณใ‚ฐใ€ไป–ใƒขใƒ‡ใƒซใจใฎใƒžใƒผใ‚ธใ€ๅ†้…ๅธƒใชใฉใ‚’่กŒใ†ๅ ดๅˆ๏ผ‰ใ€ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ณใƒผใƒ‘ใ‚นใซ็”ฑๆฅใ™ใ‚‹้ƒจๅˆ†ใฎๅ–ใ‚Šๆ‰ฑใ„ใซใคใ„ใฆใฏใ€ใคใใ‚ˆใฟใกใ‚ƒใ‚“ใ‚ณใƒผใƒ‘ใ‚นใฎๅˆฉ็”จ่ฆ็ด„ ใซๅพ“ใฃใฆใใ ใ•ใ„ใ€‚ใ“ใฎ่ฆๅฎšใฏใ€ๆดพ็”Ÿใ‚ฝใƒ•ใƒˆใ‚„ๅ†้…ๅธƒใ•ใ‚ŒใŸใƒ‡ใƒผใ‚ฟใซใ‚‚ใ‚ณใƒ”ใƒผใƒฌใƒ•ใƒˆใ•ใ‚Œใพใ™ใ€‚

Credit and Usage Conditions

Tsukuyomichan Corpus Attribution

This model was trained using voice data freely released by the free-use character "Tsukuyomichan" (ยฉ Rei Yumesaki).

When publishing this model, a demo, or an application, include the following credit in full. Please include the URL.

This software uses voice data freely released by the free-use character "Tsukuyomichan" (ยฉ Rei Yumesaki).

โ–  Tsukuyomichan Corpus (CV: Rei Yumesaki)
https://tyc.rei-yumesaki.net/material/corpus/

Short credit example for end users:

This project uses the Tsukuyomichan Corpus (CV: Rei Yumesaki).

Restrictions on Output Audio

Audio generated from this model must not be used for the following purposes:

  • Criticizing or attacking individuals. (The definition of "criticism or attack" follows the Tsukuyomichan Character License.)
  • Advocating for or against specific political positions, religions, or ideologies.
  • Publishing explicit or offensive content without appropriate age-gating.
  • Distributing or publishing in a form that permits secondary use (use as raw material) by others.

โ€ป Distribution or sale as finished creative works is permitted.

Modification and Redistribution

If you use this model itself as a base (including modification, fine-tuning, merging with other models, or redistribution), the portions derived from the Tsukuyomichan Corpus must be handled in accordance with the Tsukuyomichan Corpus Terms of Use. This requirement is copyleft and carries over to all derivative works and redistributed data.

Downloads last month
39
Safetensors
Model size
0.6B params
Tensor type
I64
ยท
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for kizuna-intelligence/tsukuyomichan-omnivoice-full-finetune

Finetuned
Qwen/Qwen3-0.6B
Finetuned
k2-fsa/OmniVoice
Finetuned
(80)
this model
Quantizations
2 models