Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/folder_based_builder/folder_based_builder.py", line 192, in _split_generators
                  raise ValueError(f"Found metadata files with different extensions: {list(metadata_ext)}")
              ValueError: Found metadata files with different extensions: ['.csv', '.jsonl']
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 65, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                         ^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Hadou Voice Dataset

Hadou Voice Dataset is a Japanese voice dataset for AI voice training, TTS experiments, voice cloning research, and related creative workflows.

This first public release contains Hadou reading ITA Corpus sentences.
Future releases may add other scripts, conversational lines, emotional speech, filler words, and additional recording styles.

日本語説明:

Hadou Voice Dataset は、AI音声学習、TTS検証、ボイスクローン研究、創作支援向けに公開する日本語音声データセットです。

現在の v0 は、ハドウ本人が ITAコーパスを読み上げた音声を、1文ごとのWAVとテキスト対応表に分割したものです。
今後、ITAコーパス以外の原稿、感情表現、会話調、フィラーワード、別収録パターンなどを追加していく予定です。

Official Links

BOOTH Description

ハドウ本人がITAコーパスを読み上げた、AI学習用の音声データです。

TTS、ボイスチェンジャー、LoRA作成、生成した音声の動画・配信・ゲーム・アプリ制作への利用などに自由に使ってください。
個人・法人・商用利用OKです。

ハドウ声のLoRA・音声モデル・プリセット等を公開する場合は、「ハドウ」の名前を表記してください。
作品内の声として使うだけならクレジットは任意です。

犯罪行為、誹謗中傷、政治・宗教活動、なりすまし、元音声データの再配布・販売・公開は禁止です。
詳しい利用規約は同梱READMEをご確認ください。

今後、感情ありや別の声質バージョンの追加も考えています。

Sample generated with Irodori-TTS V3 LoRA:
https://x.com/harddeluxe/status/2056015213842124899

English:

This is an AI training voice dataset recorded by Hadou reading the ITA Corpus.

You may freely use it for TTS, voice changers, LoRA creation, videos, livestreams, games, apps, and similar projects.
Personal, corporate, and commercial use are allowed.

If you publish a LoRA, voice model, preset, or similar item mainly intended to reproduce Hadou's voice, please credit the name "Hadou".
Credit is optional when you only use the generated voice as a voice in your own work.

Use for criminal acts, defamation or harassment, political or religious activities, impersonation, or redistribution, sale, or publication of the original voice data is prohibited.
Please see the included README for the full terms of use.

Dataset Summary

  • Speaker: Hadou
  • Language: Japanese
  • Current release: ITA Corpus reading voice dataset
  • Clips: 424
  • Total duration: approximately 38.95 minutes
  • Format: WAV
  • Metadata: metadata.csv and metadata.jsonl
  • Main use cases: TTS fine-tuning, voice cloning experiments, RVC/TTS dataset tests, AI voice research, creative prototyping

Current Contents

audio/
  emotion/
  recitation/
metadata.csv
metadata.jsonl
metadata_source.jsonl
metadata_booth.csv
metadata_simple.csv
README.txt
corpus/
dataset_report.md

Metadata

The main Hugging Face metadata file is metadata.csv.

Columns:

  • file_name: relative path to the WAV file
  • text: transcript
  • speaker: speaker name
  • style: current corpus group, such as emotion or recitation
  • id: clip id
  • source_corpus: source transcript group
  • reading: kana reading
  • duration: clip duration in seconds

The BOOTH distribution metadata is also included as metadata_booth.csv and metadata_simple.csv.

Example:

file_name,text,speaker,style,id,source_corpus,reading,duration
audio/emotion/EMOTION100_001.wav,えっ嘘でしょ。,Hadou,emotion,EMOTION100_001,emotion_transcript_utf8,エッウソデショ。,3.752

Loading

from datasets import load_dataset

dataset = load_dataset("audiofolder", data_dir=".")
print(dataset["train"][0])

If hosted on Hugging Face:

from datasets import load_dataset

dataset = load_dataset("YOUR_USER_NAME/Hadou_Voice_Dataset", data_dir=".")

Replace YOUR_USER_NAME/Hadou_Voice_Dataset with the actual repository name.

For this repository:

from datasets import load_dataset

dataset = load_dataset("hadou1225/Hadou-Voice-Dataset", data_dir=".")

Intended Use

This dataset is intended for:

  • Japanese TTS experiments
  • voice cloning / voice adaptation experiments
  • RVC and related AI voice workflows
  • dataset format testing
  • research and creative prototyping
  • non-commercial and commercial creative use, subject to the license terms below
  • LoRA, speech synthesis models, voice conversion models, voice presets, and similar models that primarily reproduce the Hadou voice, as long as the license terms are followed

Out-of-Scope / Prohibited Use

Do not use this dataset for the prohibited uses listed in LICENSE, including impersonation of Hadou, Delahado official channels, or related parties, illegal acts, harassment, political/religious solicitation, or redistribution/sale/public release of the raw voice data itself or modified copies of the raw voice data.

License / Terms

This dataset is distributed under custom terms, not CC0. See the LICENSE file for the full Japanese terms.

BOOTH distribution terms are mirrored below so that the Hugging Face release and BOOTH release do not diverge.

ハドウAI学習用音声 利用規約

内容

ITAコーパスをハドウ本人が読み上げた、AI学習用の音声データです。

台本テキスト: ITAコーパス(Public Domain / Unlicense)
https://github.com/mmorise/ita-corpus

利用について

AI学習、Text to Speechのような音声合成、ボイスチェンジャー、LoRA作成などにご利用ください。
禁止事項に該当しない限り、個人・法人・商用・表現ジャンルを問わず自由に利用できます。

ハドウの声質を主目的として再現するLoRA、音声合成モデル、声質変換モデル、プリセット等も、条件に従って公開・配布・商用利用できます。

利用にあたって、事前・事後の連絡は不要です。

クレジット

生成音声を作品内の声として利用する場合、クレジット表記は任意です。
作品の形式上、声の出演者や使用音声を表記する場面では、可能な範囲で「ハドウ」と表記していただけたら嬉しいです。

ハドウの声質を主目的として再現するLoRA、音声合成モデル、声質変換モデル、ボイス設定、プリセット等を公開・配布・販売・提供する場合は、「ハドウ」の名前を必ず表記してください。

複数話者モデル、汎用モデル、そのほかハドウの声質を主目的としない用途では、クレジット表記は不要です。

禁止事項

  • 犯罪行為、違法行為、またはそれらを助長する目的で利用する行為
  • 反社会的勢力による利用、または反社会的勢力の活動に関連する利用
  • 他者への誹謗中傷、侮辱、差別、脅迫、嫌がらせ、名誉・信用の毀損を目的とする利用
  • 政治活動、選挙活動、宗教活動、または特定の思想・信条への勧誘を目的とする利用
  • ハドウ本人、デラハド公式、または関係者になりすます行為
  • 本音声データそのもの、または本音声データを加工・分割・形式変換したデータを、再配布・販売・公開する行為
  • その他、法令または公序良俗に反する行為
  • その他、上記に準ずる不適切な行為

免責

本音声データの利用により発生した損害・トラブルについて、配布者は責任を負いません。
利用者自身の責任でご利用ください。

備考

本規約は音声データのみを対象としています。
ハドウの立ち絵などビジュアル素材が使いたい場合は、別途ご連絡ください。

話者

ハドウ(デラハド)
https://www.youtube.com/@DelaHadou
https://x.com/harddeluxe

Source Corpus

The current v0 release uses ITA Corpus text as the reading script.

This dataset is not limited to ITA Corpus in concept. Future versions may include other freely usable scripts and original lines suitable for AI character voice training.

Known Limitations

  • Current release is relatively small: approximately 39 minutes.
  • Current release is mostly controlled reading, not full conversational acting.
  • Emotional range is limited in this version.
  • Some downstream models may require resampling or metadata conversion.
  • Users should listen and inspect samples before training production models.

Related Release

A BOOTH distribution may also be available for users who prefer a simple ZIP download and Japanese instructions.

Changelog

v0

  • Initial public release.
  • 424 clips.
  • Approximately 38.95 minutes.
  • ITA Corpus reading voice dataset.
Downloads last month
39