A few errors from the Uberduck Colaboratory book

Checking Training Files wavs/1.npy|Look at that cheap squirt passin' up and down. [WARNING] does not exist. --------------------------------------------------------------------------- FileNotFoundError Traceback (most recent call last) in () 4 get_ipython().system("sed -i -- 's,.wav|,.npy|,g' {hparams.training_files}; sed -i -- 's,.wav|,.npy|,g' {hparams.validation_files}") 5 ----> 6 check_dataset(hparams) 2 frames in check_dataset(hparams) 381 print("Checking Training Files") 382 audiopaths_and_text = load_filepaths_and_text(hparams.training_files) # get split lines from training_files text file. --> 383 check_arr(audiopaths_and_text) 384 print("Checking Validation Files") 385 audiopaths_and_text = load_filepaths_and_text(hparams.validation_files) # get split lines from validation_files text file. in check_arr(filelist_arr) 375 mel_length = 1 376 if hparams.load_mel_from_disk and '.npy' in file[0]: --> 377 melspec = torch.from_numpy(np.load(file[0], allow_pickle=True)) 378 mel_length = melspec.shape[1] 379 if mel_length == 0: /usr/local/lib/python3.7/dist-packages/numpy/lib/npyio.py in load(file, mmap_mode, allow_pickle, fix_imports, encoding) 414 own_fid = False 415 else: --> 416 fid = stack.enter_context(open(os_fspath(file), "rb")) 417 own_fid = True 418 FileNotFoundError: [Errno 2] No such file or directory: 'wavs/1.npy'

------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------------------- FileNotFoundError Traceback (most recent call last) in () 7 print('cuDNN Benchmark:', hparams.cudnn_benchmark) 8 train(output_directory, log_directory, checkpoint_path, ----> 9 warm_start, n_gpus, rank, group_name, hparams, log_directory2) 6 frames /usr/local/lib/python3.7/dist-packages/torch/_utils.py in reraise(self) 423 # have message field 424 raise self.exc_type(message=msg) --> 425 raise self.exc_type(msg) 426 427 FileNotFoundError: Caught FileNotFoundError in DataLoader worker process 0. Original Traceback (most recent call last): File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/worker.py", line 287, in _worker_loop data = fetcher.fetch(index) File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch data = [self.dataset[idx] for idx in possibly_batched_index] File "/usr/local/lib/python3.7/dist-packages/torch/utils/data/_utils/fetch.py", line 44, in data = [self.dataset[idx] for idx in possibly_batched_index] File "/content/tacotron2/data_utils.py", line 61, in __getitem__ return self.get_mel_text_pair(self.audiopaths_and_text[index]) File "/content/tacotron2/data_utils.py", line 34, in get_mel_text_pair mel = self.get_mel(audiopath) File "/content/tacotron2/data_utils.py", line 49, in get_mel melspec = torch.from_numpy(np.load(filename)) File "/usr/local/lib/python3.7/dist-packages/numpy/lib/npyio.py", line 416, in load fid = stack.enter_context(open(os_fspath(file), "rb")) FileNotFoundError: [Errno 2] No such file or directory: 'wavs/111.npy'

Strangely enough, a lot of the clips were too short but somehow sounded good after training. Some of the clips were 1-6 seconds. 7 seconds max I think.

That's why the training diagram looked blocky

Yeahhh I saw that, but nice bro. It sounds really good and I'm sure it'll sound a lot better if you also made a HiFi-GAN model for it too for your own usage of the AI Model

late but did you import the files into the

wavs

folder and not another folder inside it

i tried that, thats when this error happened, then I tried the zip cuz it said use the zip if you have a ton of files (I have 154)

Hello guys, I cannot seem to find the trained model in my output path. This is what displayed when I started training the Uberduck Tacotron 2 simplified training colab

if you just started it that's because it only saves every so often

if it saved after every single epoch it'd get ratelimited i believe

By saves every so often, it saves after how many epochs?

that's the fun part: whoever made this notebook decided to make it random

it saves 15% of the time

so roughly every 6-7 epochs

Oh damn!! now I have to wait all those times. I had run it for 3 epochs. Was getting nervous and stopped it to see its outpuut but found none.

Let me wait now till like 20th epoch and check

Thank you @User

no problem

i'm working on an improved version of the notebook that lets you change this

but for now you can double-click on

Installing Tacotron2 and dependencies

, scroll down to

def save_checkpoint

, and replace

if random.random() > 0.85:

or whatever with

if True:

then it'll save every time

though i don't super recommend that because of the aforementioned rate limit

also dang that's either a massive dataset or you got a k80

Alright, let me just leave it.

I cut down the LJ speech dataset , Its what I pluggged in . And am using a T4

ah

wait why are you training on ljspeech

Just to get a hang of it