datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
twi-grapheme-unit-features
Twi Grapheme-Unit Feature Store
The training data behind ghana-pico-asr:
805 hours of Twi speech turned into log-mel features with a
grapheme-unit label for every 10 ms frame, produced by CTC forced
alignment. Publishing it means the expensive step -- aligning 805 hours
on a GPU -- does not have to be repeated to train, reproduce or extend the
model, and the same pipeline can be pointed at a new language.
This is a derived feature store, not a speech corpus: it contains 40-band… See the full description on the dataset page: https://huggingface.co/datasets/ghanaopenai/twi-grapheme-unit-features.vispeech-whisper-features
ViSpeech — Preprocessed Whisper Features (Reproducibility Release)
This repository contains the preprocessed feature sets used in the paper:
ViSpeech: A Multi-Condition Vietnamese Speech Dataset for Noise-Robust Classroom ASR — T. D. Tran et al., FISAT 2026 (Springer).
ViSpeech is a 37.62-hour multi-condition Vietnamese speech corpus built on a paired-recording design — the same speakers reading the same scripts in quiet-room and real-classroom conditions — comprising a… See the full description on the dataset page: https://huggingface.co/datasets/Hongthien06/vispeech-whisper-features.