That's a compatibility issue we can't do anything about

FileNotFoundError Traceback (most recent call last) in 115 116 # Convert to JSON --> 117 generate_json("trainfiles.txt", "trainfiles.json") 118 generate_json("valfiles.txt", "valfiles.json") 119 generate_json("allfiles.txt", "allfiles.json") in generate_json(inpath, outpath) 44 if lpath[:5] != "wavs/": 45 lpath = "wavs/" + lpath ---> 46 size = os.stat( 47 os.path.join(os.path.dirname(inpath), lpath) 48 ).st_size FileNotFoundError: [Errno 2] No such file or directory: 'wavs/44.wav'

I mean, it says it ight there

You're short a. File

I just trained my model for talknet but I don't see a zip file

There is supposed to be something that says Mikasa_Ackerman.zip but it's not there

Your drive doesn't actaully have enough room to save it

Look at your drive

Oh

Free up your trash can some

Ok

Make sure you've still got the notebook open

And dont restart the runtime

If you do you'll need to go through the steps again from step 1 but it'll load the checkpoints if you didn't just delete them trying to save it last time

You got that @Hange Zoë ?

yup @WeegeeFan1

Questino for those out there: What do you do when your doing step 6 (dataset processing) and no matter what you do, because your dataset is so big, you run out of collab gpu usage to actaully process it all the way?

I've done this and it doesn't ever contenue. It just starts from scratch again! It doesn't even save checkpoints!

What is a good length of samples to use and how many epochs would you recommend for a good voice?

total data length for a good talknet should not be less than 5 minutes from what ive heard

I've always used 500 epochs, otherwise it's too matallic sounding even with a good hifigan model

500 epochs on step 9 (spectrogram generator)? I've been under the impression that you run a high risk of over-training it after 250~ epochs

@TheRoyalRuby2000 Can You Approve My Cleveland Brown Talknet cleveland-brown

f073c20b-2e0b-4674-a4a9-2dc856e514ea

No need to send that string of characters

But There Is Already A Cleveland Brown Tacotron

Huh

And It Will Be Confusing Which One

Like I Uploaded Multiple

What is

f073c20b-2e0b-4674-a4a9-2dc856e514ea

?