i just saw the notebooks channel as well. okay for limited numbers of files? (< 100)

i mean it's the same thing you used earlier

okay - i was wondering if it was a new thing or not.

You're gonna need to do that regardless, pretty sure it wont work without modifications

I had go change some stuff to get it to work, might've been some unaddressed bugs or maybe just my specific hardware (for vocoder finetuning)

it looks like its its own thing instead of fsr

so regarding Talknet - i used an ancient version of RX (2.0) to try and deal with my background removal / isolation issues for my voice samples - music soundtrack background mostly. would Talknet do better in this role?

The Sportscasters work really well. But...wouldn't it be better to have a seperate model for the excited/shouting commentaries?

any recommends to remove a speaker from background music (not a singer)? i first used M/S (like the old fashioned vocal remover plugs do) which helped a lot, then used a really old version of Izotope RX (2.0) to wash out the rest. anyone have other tips on better isolation? i thought Talknet could do it based on the comment in the notebook channel. but just checking if anyone knows of a good tool or colab for removing background music from a speaker?

It is but it has added FSR (It's not in the release yet, but you can build from the source code)

It already had another version of FSR that just upscaled but it added Fidelity FX SR in the latest branch

there's one in #841437191073955920 right above the talknet ones but the default settings there are not optimal

try these

actually i just tried spleeter on the M channel of my vocals and it worked really well in default settings. not perfect but it rescued three lines i abandoned due to interference. very nice tool for my purposes.

100 files

got more audio files so now im training more.

out of curiosity is it tacotron or talknet

also for 100 files dang that's clean

tacatron

a

honestly if you were to add a finetuned hifi-gan i don't think people would be able to tell it's ai

I ran my model with some uberduck juice. idk if it sounds better.

It’s Gene Belcher I found out straight away

The text-to-speech voice

When will it be finished

I am a big fan of Bob’s burgers

So is my dad

He loves the show too

Yep. You got it. lol I'm not sure. I've got over 300 audio clips for my gene model. it depends on if it sounds right to other people. then I can submit it and work on the other belchers. I'm planning on giving everyone their own model so.

It sounds awesome I love it