datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ATC_combined
Dataset Card for UWB-ATCC corpus
Dataset Summary
The UWB-ATCC Corpus is provided provided by University of West Bohemia, Department of Cybernetics. The corpus contains recordings of communication between air traffic controllers and pilots. The speech is manually transcribed and labeled with the information about the speaker (pilot/controller, not the full identity of the person). The corpus is currently small (20 hours) but we plan to search for additional data next year.… See the full description on the dataset page: https://huggingface.co/datasets/Shiry/ATC_combined.ATC_combined
Dataset Card for UWB-ATCC corpus
Dataset Summary
The UWB-ATCC Corpus is provided provided by University of West Bohemia, Department of Cybernetics. The corpus contains recordings of communication between air traffic controllers and pilots. The speech is manually transcribed and labeled with the information about the speaker (pilot/controller, not the full identity of the person). The corpus is currently small (20 hours) but we plan to search for additional data next year.… See the full description on the dataset page: https://huggingface.co/datasets/Tabys/ATC_combined.Pidgin_ASR_Dataset_Combined
Naija-ASR-Corpus v1.0 (NAC-v1.0)
A Foundational Automatic Speech Recognition Corpus for Nigerian Pidgin (Naija, PCM)
📌 Dataset Summary
Naija-ASR-Corpus (NAC-v1.0) is a speech dataset derived from the Universal Dependencies Naija Spoken Corpus (UD_Naija-NSC).
The NAC Team processed the original long-form recordings by:
Segmenting the audio into sentence-level clips.
Transcribing/Aligning the text to create paired audio-text data suitable for ASR training.
This dataset… See the full description on the dataset page: https://huggingface.co/datasets/timniel/Pidgin_ASR_Dataset_Combined.Pidgin_ASR_Dataset_Combined
Naija-ASR-Corpus v1.0 (NAC-v1.0)
A Foundational Automatic Speech Recognition Corpus for Nigerian Pidgin (Naija, PCM)
📌 Dataset Summary
Naija-ASR-Corpus (NAC-v1.0) is a speech dataset derived from the Universal Dependencies Naija Spoken Corpus (UD_Naija-NSC).
The NAC Team processed the original long-form recordings by:
Segmenting the audio into sentence-level clips.
Transcribing/Aligning the text to create paired audio-text data suitable for ASR training.
This dataset… See the full description on the dataset page: https://huggingface.co/datasets/AnonXx/Pidgin_ASR_Dataset_Combined.tr-combined
Tr_combined
This is a merged speech dataset containing 221531 audio segments from 894 source datasets.
Dataset Information
Total Segments: 221531
Speakers: 2158
Languages: tr
Emotions: happy, angry, neutral, sad
Original Datasets: 894
Dataset Structure
Each example contains:
audio: Audio file (WAV format, original sampling rate preserved)
text: Transcription of the audio
speaker_id: Unique speaker identifier (made unique across all merged datasets)
language:… See the full description on the dataset page: https://huggingface.co/datasets/Codyfederer/tr-combined.somali-combined-asr-stt-dataset
Somali Combined ASR/STT Dataset
A Somali automatic speech recognition (ASR) / speech-to-text (STT) dataset combining
synthetic TTS-generated audio and other Somali speech sources, deduplicated by transcript
text and split into train/validation/test.
Dataset Summary
Language: Somali (so)
Task: Automatic Speech Recognition / Speech-to-Text
Audio format: WAV, 16 kHz mono
Total examples: 8,226 (after deduplication)
Total audio: ~6 hours
Split
Examples
Audio… See the full description on the dataset page: https://huggingface.co/datasets/IbrahimDayax/somali-combined-asr-stt-dataset.pidgin-asr-combined
Pidgin ASR Combined
A unified Nigerian Pidgin English speech-to-text dataset that combines
publicly available Pidgin ASR sources into a single train / validation /
test setup with a consistent schema. Built for fine-tuning Whisper-family
models on Nigerian Pidgin (Naija, pcm).
~8.6 hours, 4,278 clips, 10 source speakers, 16 kHz mono WAV.
Used to train michaelodafe/whisper-pidgin-v1
(21.37% WER on the test split, beating the published Wav2Vec2-XLSR-53
baseline by 8.2 pp).… See the full description on the dataset page: https://huggingface.co/datasets/michaelodafe/pidgin-asr-combined.combined_synthetic_datasets_eng_hin_engandhincodemix
Combined Synthetic Datasets (English, Hindi, Code-Mix)
Public ASR training data combining YouTube podcast VAD clips, English/Hinglish podcasts, and synthetic Hinglish entity-normalization speech.
Subsets
Config
Rows
Description
yt_video_transcript
4,100
Hindi-dominant YouTube podcast segments (VAD chunks)
vad_english
1,160
English podcast segments
vad_hindi_english
787
Hindi–English code-mixed podcast segments
synthetic_voice_stt
24,459
Synthetic… See the full description on the dataset page: https://huggingface.co/datasets/sajalmadan0909/combined_synthetic_datasets_eng_hin_engandhincodemix.tamil-combined
tamil-combined
Combined and deduplicated Tamil ASR dataset (tamil1–tamil7). Audio resampled to 16 kHz mono WAV. Columns: text, audio.
Dataset details
Field
Value
Parquet shards
61
Approx size
64.3 GB
Audio format
16 kHz mono WAV
Columns
text, audio
Load
from datasets import load_dataset, Audio
ds = load_dataset("Shubhangi7/tamil-combined", split="train")
ds = ds.cast_column("audio", Audio())