This model is english version.
If you want them to speak more descriptively in Japanese, try the cool version.
If you want them to speak emotionally in Japanese, try the sweet version
If you want them to speak whisper voice in Japanese, try the ASMR version.
If you want them to speak in Chinese, try the Chinese version
This model is a TTS (text-to-speech) model.
This is a model that has trained style_bert_vits2 with my own voice data.
style_bert_vits2 is a speech generation model.
Compared to previous models, it is possible to generate highly accurate and natural speech.
Since the training data is only the voice of the researcher who created the model,
The license is the same as style_bert_vits2
You can use it freely and free of charge, regardless of whether it is commercial or non-commercial.
モデルを使うときのお約束/limitation
〇できること/What you can do
成果物の加工 Processing of deliverables
成果物の商用利用 Commercial use of deliverables
成果物の学習素材としての利用 Use of deliverables as learning materials
X(Twitter)や説明文でこのモデルを使ったことを書いてもらえると作者が喜びます。(必須ではありません)
If you write that you are using this model, I will be glad!
モデルの使い方/how to use (コードはgoogle colab用です。 For google colab)
2通りの使用方法があります。必要に応じて選択してください。There are 2 ways to use model.
1.style-bert-vits2のアプリを使ってボイスを生成する/to use style-bert-vits2 app
①Style-Bert-VITS2 インストール先の Style-Bert-VITS2/model_assets/rikka_botan/ フォルダに config.json, safetensors, style_vectors.npy の 3ファイルを置きます。
Put 3 files on Style-Bert-VITS2/model_assets/rikka_botan/ folder
以下のプログラムで自動的に保存できます。By using this program, we can save files.
1# At first, we will install the required libraries2!git clone https://github.com/litagin02/Style-Bert-VITS2.git
3%cd Style-Bert-VITS2/4!pip install -r requirements.txt
5!pip install style-bert-vits2 --no-build-isolation # To avoid bugs67# load Japanese bert model8from style_bert_vits2.nlp import bert_models
9from style_bert_vits2.constants import Languages
1011bert_models.load_model(Languages.EN,"microsoft/deberta-v3-large")12bert_models.load_tokenizer(Languages.EN,"microsoft/deberta-v3-large")1314# save model files to model_assets dir15from pathlib import Path
16from huggingface_hub import hf_hub_download
1718model_file ="rikka_botan_english.safetensors"19config_file ="config.json"20style_file ="style_vectors.npy"2122forfilein[model_file, config_file, style_file]:23print(file)24 hf_hub_download(25"RikkaBotan/style_bert_vits2_english_original",26file,27 local_dir="model_assets"28)293031# By using saved model, we will test text-to-speech demo32from style_bert_vits2.tts_model import TTSModel
3334assets_root = Path("model_assets")3536model = TTSModel(37 model_path=assets_root / model_file,38 config_path=assets_root / config_file,39 style_vec_path=assets_root / style_file,40 device="cuda"# If you cannot use cuda, please input cpu41)4243# Please input the English text44from IPython.display import Audio, display
4546sr, audio = model.infer(text="please write text here")47display(Audio(audio, rate=sr))
謝辞/Acknowledgments
style-bert-vits2を開発してくださったlitaginさんに感謝いたします
また、本モデルは多くの研究者さんの努力の上にできています。先人たちの努力に深く感謝します。
We would like to thank Mr./Ms. litagin for developing style-bert-vits2.
Additionally, this model was created based on the efforts of many developers. We are deeply grateful for the efforts of our predecessors.