A fix for a lack of high pitched audio?

[karmic fiber]

I've got a very clean dataset of someone talking, but I have a complete lack of high pitched audio and it kinda causes the model to spazz out when it comes to singing something like Viva la vida. Is there some way to fix that?

[opal patrol]

If you can't get high pitched audio of that person, you could try to change the pitch of the vocals

-12 specifically is an octave deeper and I'd recommend using that so it's in tune

[karmic fiber]

Would you recommend putting that result into a dataset or no?

[opal patrol]

no, I'd just recommend changing the pitch before inference

in the dataset it may cause issues (and it's deeper too)

[karmic fiber]

alright

thanks

[opal patrol]

You're welcome :)

Mark this post as solved btw