The Pidgin dataset for Text to Speech is Wrong
#23
by voicedata - opened
I played some of the samples in the pidgin dataset; the audio responses are different from the text associated with them. Kindly look into it
e.g speaker_id: 3
text: One-tenth of 1 million na 100 thousand.
text:
Clear sunny Wednesday, na better weather to wash cloth keep for sun