datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
OpenBible_Swahili_book_splitcommon_voice_13_0_bn_multi_splittw_parliament_split
Parliament
Parliament is an OpenFormosa Traditional Chinese speech dataset derived from the
zh_tw split of disco-eth/WorldSpeech.
It contains audio clips and human transcripts from Taiwan Legislative Yuan IVOD
parliamentary proceedings.
This release keeps only rows that passed the Taiwan-OmniData / FineWeb2-style
text filtering pipeline. Audio is preserved from the upstream dataset and cast
as a Hugging Face Audio(sampling_rate=24000) feature.
Dataset Summary… See the full description on the dataset page: https://huggingface.co/datasets/NickWeng/tw_parliament_split.japanese-anime-speech-v2-split-150k
japanese-anime-speech-v2-split-150k
joujiboi/japanese-anime-speech-v2 的 150,000 筆隨機子集，已切好 train / test。
split
rows
train
135,000
test
15,000
total
150,000
Columns
audio — 16 kHz mp3，與原始資料完全相同（未重新編碼）
sentence — 轉錄文字（原始欄位名為 transcription）
How it was built
來源的 sfw（271,788 筆）與 nsfw（20,849 筆）兩個 split 都有使用，並依原始比例分配名額
（sfw 139,313 / nsfw 10,687）。
在每個 split 的全部列上做無放回均勻抽樣，因此每筆資料被選中的機率相同。
抽出後整體打亂，前 15,000 筆為 test，其餘為 train，train / test… See the full description on the dataset page: https://huggingface.co/datasets/hhim8826/japanese-anime-speech-v2-split-150k.Emilia-dataset-french-splitghana-female-twi-speech-asr-8word-splits
This dataset is shared under CC BY-NC 4.0, which means you are free to use, share, and adapt it for non-commercial research and educational purposes with attribution. You can read the full license at https://creativecommons.org/licenses/by-nc/4.0/.
Twi 8-Word Speech Segments
51139 speech-text pairs split from 30-min recordings.
Processing pipeline
Source audio from ghananlpcommunity/ghana-female-twi-tts-full-length
Full-file CTC forced alignment (MMS-300M) for… See the full description on the dataset page: https://huggingface.co/datasets/ghanaopenai/ghana-female-twi-speech-asr-8word-splits.earnings22_robust_splitfrom datasets import load_dataset, DatasetDict
ds = load_dataset("anton-l/earnings22_robust", split="test")
print(ds)
print("\n", "Split to ==>", "\n")
# split train 90%/ dev 5% / test 5%
# split twice and combine
train_devtest = ds.train_test_split(shuffle=True, seed=1, test_size=0.1)
dev_test = train_devtest['test'].train_test_split(shuffle=True, seed=1, test_size=0.5)
ds_train_dev_test = DatasetDict({'train': train_devtest['train'], 'validation': dev_test['train'], 'test':… See the full description on the dataset page: https://huggingface.co/datasets/sanchit-gandhi/earnings22_robust_split.wikipedia_asr_splittedSynthetic dataset based on highly specialised texts from Wikipedia. This version is splitted by category. Voiced using Yandex SpeechKit with random voices, roles and speech rate.
Can be used to evaluate ASR models not trained on given domains and to identify areas that the model does not handle well.
Non-splitted version can be found here: https://huggingface.co/datasets/rmndrnts/wikipedia_asr
earnings22_splitWe partition the earnings22 dataset at https://huggingface.co/datasets/anton-l/earnings22_baseline_5_gram by source_id:
Validation: 4420696 4448760 4461799 4469836 4473238 4482110
Test: 4432298 4450488 4470290 4479741 4483338 4485244
Train: remainder
Official script for processing these splits will be released shortly.
MSPP_WAV_speaker_splitvlsp2020_vinbig_100h_splitsh_val_splitcommon-voice-21.0-2025-03-14-zh-CN-splitghana-female-twi-8sec-splits
This dataset is shared under CC BY-NC 4.0, which means you are free to use, share, and adapt it for non-commercial research and educational purposes with attribution. You can read the full license at https://creativecommons.org/licenses/by-nc/4.0/.
Twi 8-Word Speech Segments
25951 speech-text pairs split from 30-min recordings.
Processing pipeline
Source audio from ghananlpcommunity/ghana-female-twi-tts-full-length
Full-file CTC forced alignment (MMS-300M) for… See the full description on the dataset page: https://huggingface.co/datasets/ghananlpcommunity/ghana-female-twi-8sec-splits.lena_dataset_splittedSubset-splitted version of a synthetic dataset based on specific domains. The textual basis was picked up by Elena Bruches.
The non-splitted version can be found here: https://huggingface.co/datasets/rmndrnts/lena_dataset
earnings22_split_resampledWe partition the earnings22 dataset at https://huggingface.co/datasets/anton-l/earnings22_baseline_5_gram by source_id:
Validation: 4420696 4448760 4461799 4469836 4473238 4482110
Test: 4432298 4450488 4470290 4479741 4483338 4485244
Train: remainder
Official script for processing these splits will be released shortly.
ghana-female-twi-asr-16word-splits
This dataset is shared under CC BY-NC 4.0, which means you are free to use, share, and adapt it for non-commercial research and educational purposes with attribution. You can read the full license at https://creativecommons.org/licenses/by-nc/4.0/.
Twi 16-Word Speech Segments
25951 speech-text pairs split from 30-min recordings.
Processing pipeline
Source audio from ghananlpcommunity/ghana-female-twi-tts-full-length
Full-file CTC forced alignment (MMS-300M) for… See the full description on the dataset page: https://huggingface.co/datasets/ghananlpcommunity/ghana-female-twi-asr-16word-splits.resampled_16KHrz_vctk_speakers_splitluganda-english-cleaned-v1-splitiemocap_audio_text_splitted
Dataset Card for "iemocap_audio_text_splitted"
More Information needed
naija-voices-hausa-split_0-1Clean_One_Speaker_STT_Split_EN_AR
Clean_One_Speaker_STT_Split_EN_AR
Mixed STT dataset — Saudi dialectal Arabic + Modern Standard Arabic + English —
with train/validation/test splits, built for fine-tuning multilingual ASR (e.g. Whisper)
without catastrophic forgetting of English.
Composition
Dialectal Arabic (~72%) — Sebssihakim/Clean_One_Speaker_SADA_Split
(cleaned single-speaker SADA; SDAIA / Saudi Broadcasting Authority, CC BY-NC-SA 4.0)
MSA (~8%) — FLEURS ar_eg, full official splits (CC-BY;… See the full description on the dataset page: https://huggingface.co/datasets/Sebssihakim/Clean_One_Speaker_STT_Split_EN_AR.S2T_Split_NoRom_phase2uwb_atcosim_split_by_person_6_2_2
UWB-ATCC + ATCOSIM — combined ATC ASR dataset (ATCOSIM 6-2-2, gender-balanced)
A combined air-traffic-control ASR dataset built from Jzuluaga/uwb_atcc and Jzuluaga/atcosim_corpus, re-split with leakage-free (group-disjoint) train/validation/test splits. UWB-ATCC is split recording-disjoint (80/10/10) and ATCOSIM is split speaker-disjoint and gender-balanced, putting 1 male + 1 female speaker in each of validation and test (6 train / 2 / 2 speakers), with a fixed seed (42) and 16… See the full description on the dataset page: https://huggingface.co/datasets/mmonzel/uwb_atcosim_split_by_person_6_2_2.naija-voices-igbo-split_1-1naija-voices-yoruba-split_2-3uwb_atcosim_split_by_person_8-1-1
UWB-ATCC + ATCOSIM — combined ATC ASR dataset (ATCOSIM 8-1-1)
A combined air-traffic-control ASR dataset built from Jzuluaga/uwb_atcc and Jzuluaga/atcosim_corpus, re-split with leakage-free (group-disjoint) train/validation/test splits. UWB-ATCC is split recording-disjoint (80/10/10) and ATCOSIM is split speaker-disjoint into 8 train / 1 validation / 1 test speakers, with a fixed seed (42) and 16 kHz audio. Columns: id, audio, text, segment_start_time, segment_end_time… See the full description on the dataset page: https://huggingface.co/datasets/mmonzel/uwb_atcosim_split_by_person_8-1-1.japanese-anime-speech-v2-splitdataset split from joujiboi/japanese-anime-speech-v2
naija-voices-hausa-split_0-6naija-voices-yoruba-split_0-6