A 15‑hour single‑speaker Japanese dysarthric speech corpus featuring parallel, phonemically hard, and daily‑use utterances for ASR research, provided in 44.1 kHz RIFF WAV format.
Description in other languages
Statistics
Statistics as of Nov 16, 2026
Hugging Face downloads (last 30 days)
164
Hugging Face likes
2
Listed
Nov 3, 2026
Popular in the same category
- fgo_voices_jpdeepghs10kJP Voice‑Text Dataset for FGO Waifus: a 30,800‑record, 66.4‑hour Japanese audio collection of single‑voice‑actor character lines (≈7.76 s each) useful for ASR/ASV fine‑tuning and evaluation.
- japanese-anime-speechjoujiboi946Japanese Anime Speech Dataset offers 73,004 audio‑text pairs (110 hours total, evolving from V1 to V5) to enhance ASR models such as OpenAI’s Whisper, available under an open license for all uses with credit appreciated.
- makise-kurisu-vn-voicelineszhonglongbao847Transcribed Makise Kurisu VN dialogue from a video using Whisper Large‑V2, split into clips with pydub; uncleaned text is intended for TTS model training and does not belong to the author.
- Galgame_Speech_ASR_16kHzlitagin395Galgame_Speech_ASR_16kHz is a 16 kHz ASR dataset with 3.75 million pairs (≈5,354 h), derived from Galgame_Dataset, released under GPL v3.0 with commercial use prohibited and requiring any trained models to be open‑source (citation optional).
- azurlane_voices_jpdeepghs332A Japanese voice‑text dataset for Azur Lane characters, comprising 30 160 recordings (≈75.8 h) from single‑actor voices, useful for fine‑tuning or evaluating ASR/ASV models.
- Japanese-Eroge-VoiceNandemoGHS325A 409‑hour Japanese eroge voice dataset processed with 2‑pass loudnorm (‑23 LUFS, ‑1 dB peak, 11 LRA), transcribed by litagin/anime-whisper, anonymized, stored as WebDataset (FLAC, JSON, TXT), largely featuring female voices with potential AI transcription errors, and MIT‑licensed for academic research.
Resources in similar research fields
- arknights_voices_jpdeepghs285JP Voice‑Text Dataset for Arknights Waifus: 10,905 Japanese voice clips (26.3 h total, avg 8.7 s) from single‑actor characters, useful for fine‑tuning or evaluating ASR/ASV models.Shared: Speech recognition・Corpora & datasets
- asr-ja-benchthird-intelligence270This repository documents the corrected transcript set for the Japanese ASR benchmark “asr‑ja‑bench,” detailing sample inclusion from JSUT, JVNV, TEDxJP‑10K, FLEURS, and SpeechBSD, the applied label fixes, hash checks, and versioned results (Haas‑ja‑v1.0), while noting that no audio or transcripts are redistributed due to licensing constraints.Shared: Speech recognition・Corpora & datasets
TEDxJP-10Klaboroai28TEDxJP-10K ASR Evaluation DatasetGitHubShared: Speech recognition・Corpora & datasets- girlsfrontline_voices_jpdeepghs176A Japanese voice‑text corpus of 12 508 recordings (≈20.9 h total) from Girls Front Line “waifus,” each file featuring a single voice actor, designed for fine‑tuning and evaluating ASR/ASV models.Shared: Speech recognition・Corpora & datasets
- wav2vec2-large-xlsr-53-japanesejonatasgrosman16MJapanese wav2vec‑2 XLSR‑53 fine‑tuned on Common Voice 6.1, CSS10, and JSUT, requiring 16 kHz audio and usable via HuggingSound or HuggingFace pipelines.HF modelShared: Speech recognition
- KakologArchivesKakologArchives74kAggregates over 150 GB of NicoNico Live’s comment logs from 2009‑2024—including pre‑transition, post‑transition, and real‑time NX‑Jikkyo captures—providing an API for easy retrieval of historical TV‑broadcast discussions.Shared: Corpora & datasets