datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
round2-oss-matched
round2-oss-matched — 第二轮 4 组实验数据（每组 10 节点，共 40）
代码：repo 分支 claude/round2-matched-compute（先 git fetch origin && git merge origin/claude/round2-matched-compute）。
目录：
exp0_20b/node00..04/pool.jsonl # 实验 0:shard-05 修复重跑(20B)
exp0_120b/node00..04/pool.jsonl # 实验 0:同上(120B)
exp1_20b/node00..09/{seeds,budgets,pool}.jsonl # 实验 1:20B token 对齐独立采样
exp2_120b/node00..09/{seeds,budgets,pool}.jsonl # 实验 2:120B 同上
exp3_120b/node00..09/{ck_nonsat/,nonsat_seeds,budgets,pool… See the full description on the dataset page: https://huggingface.co/datasets/tts-sft/round2-oss-matched.round4-independent
ROUND 4 — combined r2+r3 pool, oracle-fb + independent, WITH REASONING RETENTION
Cut 2026-08-12. The first generation round whose outputs keep the gpt-oss
analysis channel (reasoning) on disk — see tts-sft/docs/REASONING_RETENTION.md.
Rounds 2–3 saved only the harmony final channel; their reasoning is unrecoverable.
What round 4 is
The combined pool — round-2 rerun pool (4,322 problems, apps-*/cc-*) +
round-3 pool (2,833 problems, cc3-*), zero id overlap, 7,155… See the full description on the dataset page: https://huggingface.co/datasets/tts-sft/round4-independent.rgad-crosslingual-tts-10h
RGAD Cross-Lingual TTS 10h
This is a 10-hour cross-lingual TTS dataset for prompt-conditioned Chinese TTS fine-tuning.
Format
The dataset contains:
train.jsonl
dev.jsonl
metadata.csv
audio/prompts/*.wav
audio/targets/*.wav
Each JSONL row has this format:
{"id":"sample_000001","prompt_wav":"audio/prompts/sample_000001.wav","target_wav":"audio/targets/sample_000001.wav","text":"中文目标文本。","prompt_language":"en-US","target_language":"zh-CN"… See the full description on the dataset page: https://huggingface.co/datasets/isabeth/rgad-crosslingual-tts-10h.tts_llmsr_data
tts_llmsr_data
naijavoices_dataset_85_hours_tts_bestFull dataset re-upload with more statistics as well as filtering scripts that give the top x files or best x hours for tts based and calculations from acoustic metrics\color{Blue}{\large \textbf{Full dataset re-upload with more statistics as well as filtering scripts that give the top x files or best x hours for tts based and calculations from acoustic… See the full description on the dataset page: https://huggingface.co/datasets/David-A-Amoo/naijavoices_dataset_85_hours_tts_best.Xijinping-TTS-Voicebank
习近平音源
所有声音资料来自公开影像，属于公有领域目前有 1h30m 的截取后声音，足够进行 Fine-tuning
Usage
按句截取
python -m pip install -r requirement.txt
python split.py
新增声音资料后，使用 Whisper 产生带有时间标记的 JSON 档，并手动复制到 ./voice/[FILE].json
export OPENAI_API_KEY="API_KEY_HERE"
python whisper.py ./[FILE].[AUDIO_EXTENSION]
产生 Bert-VITS2 微调所需的 esd.list 档案
python index_to_list.py
yodas-ttstts-embed-dataset-aime83-23Indic_New_dataset_TTS
Indic TTS Dataset Hub (Mozilla)
Validated audio–text pairs for multiple Indic languages from Mozilla Common Voice.
Select the language from the Subset dropdown in the Dataset Viewer.
Columns
audio: WAV audio clip (16kHz)
text: transcription
duration: length in seconds
speaking_rate: characters per second
tts-embed-dataset-amc23sopho-poetry-tts-data
sopho-poetry-tts-data
Training data for the SuFei on-device Chinese-poetry TTS pipeline
(sopho-poetry-tts-train).
Two co-equal lineages, one per teacher model — both active (the FS2 set is
the provenance of production v9 and stays reusable).
Lineage
Teacher
Trained
Poems
cosyvoice3/
Fun-CosyVoice3-0.5B (Apache-2.0)
m3_v6
1023
paddlespeech_fs2/
PaddleSpeech FS2 CSMSC (Apache-2.0)
v9
320
Layout
poems.jsonl # shared source text, keyed… See the full description on the dataset page: https://huggingface.co/datasets/Sopho/sopho-poetry-tts-data.nonverbal-tts-filtered-tokens
MintTTS Pre-tokenized Audio Tokens
Pre-extracted audio codec tokens for TTS training.
Source
Dataset: somu9/nonverbal-tts-filtered
Codec: MOSS-Audio-Tokenizer-Nano
Codec sample rate: 48,000 Hz (stereo)
Frame rate: 12.5 Hz (1 frame = 80ms)
Stats
Metric
Value
Total samples
6,082
Total audio hours
15.6h
Codebooks
16
Avg frames/sample
115.3
Avg duration
9.2s
Format
JSONL file (manifest.jsonl) where each line is:
{
"text": "The… See the full description on the dataset page: https://huggingface.co/datasets/somu9/nonverbal-tts-filtered-tokens.all15_speaker_deduped_tts_train_clone_pairs_raw
All-15 Speaker-Deduped TTS Train Clone Pairs Raw
This dataset contains raw metadata rows for speaker-deduped TTS voice-clone training pairs. It does not contain audio bytes. Rows point back to source audio records and include reference/target metadata, language, dataset, tier, and precomputed speaker-similarity fields from the mining pipeline.
Contents
data/train/distinct_speaker_clone_pair_plan.jsonl.gz: all survivor rows.
data/by_dataset/*.jsonl.gz: the same… See the full description on the dataset page: https://huggingface.co/datasets/instinct-org/all15_speaker_deduped_tts_train_clone_pairs_raw.tts-embed-dataset-aime2024tts-embed-dataset-math500tts-embed-dataset-gpqa-diamondtts-embed-dataset-aime2025tts-token-shimmer-stdSNAC Decoder - decoder.py
Parses annotated_data or token list and decodes to WAV audio file
Usage:
# From annotated_data string
python decoder.py --annotated "<|START_OF_USER|>...<|end|>" --output audio.wav
# From token list
python decoder.py --tokens "[123, 456, 789, ...]" --output audio.wav
# From JSON file
python decoder.py --json sample.json --output audio.wav
"""
import re
import json
import torch
import torchaudio
import argparse
from snac import SNAC… See the full description on the dataset page: https://huggingface.co/datasets/humair025/tts-token-shimmer-std.