there are a bunch of download options though i say pick either the first (newest) or one of the last two (oldest but could work better for weakest hardware)

I think I made most of mine on my phone. More annoying too imo

Having the chromebook tho at least makes it somewhat better, even with all the annoying reboots

oh chromebook

Minor thing about the pipeline notebook, what is the f16_run thing? Is that just an unused param?

theoretically it trains using 16-bit floats instead of 32-bit

which is an eency bit less precise but faster

So exporting the wavs as 16 bit is useless if it trains as 32 anyway?

nono this is for the inner neurons not the wavs

oh

--------------------------------------------------------------------------- FileNotFoundError Traceback (most recent call last) /usr/local/lib/python3.7/dist-packages/hydra/_internal/instantiate/_instantiate2.py in _call_target(target, partial, args, kwargs, full_key) 91 try: ---> 92 return _target_(*args, **kwargs) 93 except Exception as e: 10 frames FileNotFoundError: [Errno 2] No such file or directory: '/content/drive/My Drive/talknet/USMC/durations.pt' The above exception was the direct cause of the following exception: InstantiationException Traceback (most recent call last) /usr/local/lib/python3.7/dist-packages/hydra/_internal/instantiate/_instantiate2.py in _call_target(target, partial, args, kwargs, full_key) 95 if full_key: 96 msg += f"\nfull_key: {full_key}" ---> 97 raise InstantiationException(msg) from e 98 99 InstantiationException: Error in call to target 'nemo.collections.asr.data.audio_to_text.AudioToCharWithDursF0Dataset': FileNotFoundError(2, 'No such file or directory') full_key: model.train_ds.dataset

i keep getting this error.

Lol guys here is a TalkNet model I trained yesterday. I'm pretty proud of this actually.

whose voice is that?

joe

i having a very similar issue to this, i by-passed it by disabling the checkpoints from the cfg dictionary

but i cannot proceed because the models are not saved, the code is very very convoluted, did anyone manage to do it?

@hecko Since I can't post images in #910532717819596801

get this man image perms

Somehow, I've made a metronome model out of Edward M. Favor

Kinda reminds me of the recording sound in the Photo Booth app

🤔

omg ur right

just found another vid of him from 1903 that's (hopefully) good enough so back to the data

Bro how? Could make some sick lowfi hiphop with that snare sample lmao.

Forgot the input, was something about Jacob Collier.

Kinda appropriate, don't you think?