datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
azerbaijani_asr
Azerbaijani ASR Dataset
Dataset Description
This dataset contains Azerbaijani speech data for Automatic Speech Recognition (ASR) tasks.
Dataset Summary
Language: Azerbaijani (az)
Task: Automatic Speech Recognition
Total Duration: ~328 hours
Total Samples: ~345,643 audio-text pairs
Audio Format: WAV, 16kHz sampling rate
License: CC-BY-4.0
Dataset Structure
Each audio segment is specially numbered so that you can merge them if you… See the full description on the dataset page: https://huggingface.co/datasets/LocalDoc/azerbaijani_asr.localaiiaasr
Emergency Room FAQ ASR Evaluation Set
Small audio evaluation set for testing automatic speech recognition (ASR) on
emergency-room FAQ questions, recorded in Taiwanese Hokkien (nan), Mandarin
(cmn), and code-switched Hokkien/Mandarin (mixed). Most questions are
spoken in both Hokkien and Mandarin, so most transcriptions have a matching
pair of audio clips; a small subset also has a mixed-language clip.
The questions come from eval/FAQ_dataset in the
local_aiia project, grouped… See the full description on the dataset page: https://huggingface.co/datasets/TonyFANgr/localaiiaasr.fleurs-azerbaijani-asr
FLEURS Azerbaijani ASR Benchmark
Azerbaijani (az_az) subset of FLEURS,
reformatted for ASR benchmarking and fine-tuning.
Source
Based on FLEURS dataset by Google (Conneau et al., 2022).
Licensed under CC-BY-4.0.
Structure
Split
Samples
Duration
train
2656
9.28h
dev
400
1.35h
test
921
3.23h
Fields
audio — 16kHz mono WAV
sentence — transcription (original casing and punctuation)
sentence_normalized — normalized (lowercase, no… See the full description on the dataset page: https://huggingface.co/datasets/LocalDoc/fleurs-azerbaijani-asr.Bangali_local_dialect_ASR_HF_Dataset
BanglaMix — Code-Switching ASR in Bangladeshi Regional Dialects
BanglaMix is a speech dataset for Automatic Speech Recognition (ASR) on dialectal Bangladeshi Bengali mixed with English (code-switching). It covers 15 regional dialects and the natural Bengali–English code-switching common in informal Bangladeshi speech — a setting not covered by existing Bengali corpora, which address either dialects or code-switching, never both.
Clips
41,499 transcribed audio clips… See the full description on the dataset page: https://huggingface.co/datasets/niloycste68/Bangali_local_dialect_ASR_HF_Dataset.localingua_pt-brtranscriptions unverified! known to contain mistakes/noise
locallingua_ptRecordings from Portugal downloaded from https://localingual.com
localingua_africa_pt