datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
index-tts
Unofficial IndexTTS v2 Training Repo
Loop and trainer implemented using Codex CLI and guided prompts
Train new languages by extending existing tokenizer
tools\tokenizer\train_bpe.py and tools\tokenizer\extend_bpe.py
Preprocess data to extract speaker embeddings for timbre, emotion, text, and mel tokens
tools\preprocess_data.py and tools\preprocess_multiproc.py (multiproc is an attempt to make it run faster, there are issues with it though crashing)
Create prompt/target pairs… See the full description on the dataset page: https://huggingface.co/datasets/echodict/index-tts.odia-indextts2-processedodia-indextts2-processedMeow_Chan_IndexTTS2_Black_Butler
Lu_Yin_IndexTTS2_Boogiepop_Series_50_segments
MaZhengYang_IndexTTS2_Shakespeare_Audio
refer audio
Lu_Yin_IndexTTS2_Irregular_at_Magic_High_School_50_segments
Xiang_IndexTTS2_Ad_Audioindex-tts-2-examplesindex-ttsIndextts_expansion
IndexTTS2 German Training Data
Pre-processed German speech data for fine-tuning IndexTTS2.
Contents
File
Description
audio/
18,597 WAV files (24kHz)
manifest.jsonl
IndexTTS-compatible manifest
german_bpe.model
SentencePiece tokenizer
german_bpe.vocab
Vocabulary file
Stats
Language: German (DE)
Duration: 50 hours
Samples: 18,597
Source: Emilia-YODAS dataset
Quality filter: DNSMOS ≥ 3.0
Sample rate: 24kHz
Usage
# Clone the… See the full description on the dataset page: https://huggingface.co/datasets/bamselarsen/Indextts_expansion.IndexTTS_sample_0Lu_Yin_IndexTTS2_Sugar_Apple_Fairy_Tale_segments
mapalo-indextts-metadata1-v1IndexTTS-nano-codecOrigin_IndexTTS_CuteBoy_Spoken_Datasetgenshin_voices_5.6_sample_1_IndexTTSfrom datasets import load_dataset
ds = load_dataset("svjack/genshin_voices_5.6_sample_1_IndexTTS")
dss = ds["train"].filter(lambda x: x["speaker"] == "Eula" )
audio_decoder = dss[0]["reference_audio"]
import os
# 获取 _hf_encoded 属性
hf_encoded = audio_decoder._hf_encoded
with open(hf_encoded["path"], 'wb') as f:
f.write(hf_encoded["bytes"])
Xiang_IndexTTS2_Rounds_of_Phantom_Thief_Flanure
Scaramouche_IndexTTS2_Ad_AudioDont_be_your_lover_IndexTTS_CuteBoy_Spoken_DatasetIndexTTS_sample_1