Do you use both? Do they produce different results?

i haven't done a direct comparison yet

you'll probably want to use the regular notebook for now, until pipeline models become submittable

hi, whenever i start training in the tacotron notebook it stops after 44-45 minutes, even after running the anti-disconnect cell

it reaches around 1.3

it shouldn't stop that early, maybe after 3 hours but not this your internet might be unstable

and 1.3?! uhh what does the progress bar look like, is it long or very short

actually nvm now i reached 0.7 in under 20 minutes lol

with a t4

Hm

that's very slow

especially with a t4

i mean 0.07 sorry

i have 50 or so wav clips

0.07 should be enough training

idk i think i left it sitting for longer than i shouldve

i think i shouldve stopped at sub 0.09

it can overtrain if left for too long, but its clear when that happens

What does this error mean?

Exception ignored in: <function _MultiProcessingDataLoaderIter.__del__ at 0x7efc7517def0>
Traceback (most recent call last):
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/dataloader.py", line 1328, in __del__
    self._shutdown_workers()
  File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/dataloader.py", line 1320, in _shutdown_workers
    if w.is_alive():
  File "/usr/lib/python3.7/multiprocessing/process.py", line 151, in is_alive
    assert self._parent_pid == os.getpid(), 'can only test a child process'
AssertionError: can only test a child process

What notebook is that in

Recent arpabet one in #841437191073955920

Ok i've never seen that error before

Thanks to JJ03, I finally got the PCAVS notebook working.

Pog!

What's the best notebook for natural sounding speech synthesis?

TalkNet

@PUMPKINEATER Link please

You consider this "natural"

That's impressive noice