datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
reazonspeech-v2-quality-index
ReazonSpeech v2 Quality Index
Quality metadata for 21,932,215 ReazonSpeech v2 utterances.
It joins the following two source analyses by exact audio path:
ayousanz/reazon-speech-v2-all-speechMOS-analyze/audio_analysis_results_speechMOS.json
ayousanz/reazon-speech-v2-all-WAND-SNR-analyze/reazonspeech-all-wada-snr.json
The source repositories are not modified and this repository does not contain
the source audio.
Validation
Check
Count
SpeechMOS rows
21… See the full description on the dataset page: https://huggingface.co/datasets/ayousanz/reazonspeech-v2-quality-index.en-podcast-1-quality-50
Podcast quality scores — 50 recordings
First 50 recordings from Vyvo/en-podcast-1.
Original audio and sample rates are preserved.
Column
Meaning
nisqua
MLX NISQA overall MOS
dnsmos
MLX DNSMOS P.835 OVRL
quality_score
New combined score, 0–100
source_quality_score
Original dataset score
quality_score = clip(25 * ((nisqua + dnsmos) / 2 - 1), 0, 100)
All 50 scores were checked against the reference implementations. The combined
score is not human-calibrated.… See the full description on the dataset page: https://huggingface.co/datasets/Aynursusuz/en-podcast-1-quality-50.sbpn-diarized-quality-mp3-20260721
sbpn-diarized-quality-mp3-20260721
This gated dataset contains 2,242 MP3 speech chunks from
30 recordings. Access requires manual approval by the
repository owner.
Audio selection
Every complete source recording was separated once with HTDemucs.
Full vocal stems were stored remotely as 48 kHz mono 96k MP3;
no per-chunk source separation was performed.
Chunks with sustained music or background beats were cut from the saved full
vocal MP3. Other chunks were cut… See the full description on the dataset page: https://huggingface.co/datasets/Kppwdfgu1/sbpn-diarized-quality-mp3-20260721.