Model quality on cca 200hour of audio

#8
by Viewegger - opened

Hello,

as a non-spanish speaker may I ask you how good is the model when you use only 200 hours of audio data?

Hello, the model is decent, but not perfect. I'd like to retrain the model with more and better data and more VRAM (larger batch size)

thank you!

Viewegger changed discussion status to closed

Sign up or log in to comment