This corpus is a phoneme‑level derivative of the original TIMIT Acoustic‑Phonetic Continuous Speech Corpus.Each entry pairs a 1‑second waveform excerpt with a single phoneme label taken from the 61‑phone TIMIT inventory
(the mapping to 39‑phone and broad‑class sets is also provided).
This version is designed for quick prototyping of phoneme classifiers or probing acoustic representations.
Automatic… See the full description on the dataset page:
https://huggingface.co/datasets/IParraMartin/TIMITPhones.
