Common Voices is an open-source dataset of speech recordings created by
Mozilla to improve speech recognition technologies.
It consists of crowdsourced voice samples in multiple languages, contributed by volunteers worldwide.
Challenges: The raw dataset included numerous recordings with incorrect transcriptions
or those requiring adjustments, such as sampling rate modifications, conversion to .wav format, and other refinements
essential… See the full description on the dataset page:
https://huggingface.co/datasets/ionut-visan/CommonVoicesDelta21_ro.
