Uberduck

The wavsample is not changing the bitrate of my wav files.

There's a colab to fix that in <#841437191073955920> under the `Other notebooks` section.

https://colab.research.google.com/drive/100uSy4yDgr118RfuSojVumI23nM4OBkY?usp=sharing#scrollTo=IXbqIuGG8ai

```js
RuntimeError                              Traceback (most recent call last)
<ipython-input-21-37fdb9b04f8a> in <module>
      3 print("Generating mels")
      4 if generate_mels:
----> 5     create_mels()
      6 
      7 print("Checking for missing files")

3 frames
/content/tacotron2/stft.py in transform(self, input_data)
     82 
     83         # similar to librosa, reflect-pad the input
---> 84         input_data = input_data.view(num_batches, 1, num_samples)
     85         input_data = F.pad(
     86             input_data.unsqueeze(1),

RuntimeError: shape '[1, 1, 21654]' is invalid for input of size 43308```

i don't think it fixes anything by itself, it just checks

the pipeline notebook has a fixer but you're best off learning how to fix it yourself

namely
- drag it into audacity
- set sample rate in the bottom left to 22050
- tracks→ mix → mix stereo down to mono
- export as wav (16-bit)

but is there a way to do this at once with all the audios?

but instead of normal export do export multiple

```js
Generating mels
Generating Mels
37%
11/30 [00:00<00:00, 32.64it/s]
/content/tacotron2/utils.py:14: WavFileWarning: Chunk (non-data) not understood, skipping it.
  sampling_rate, data = read(full_path)
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-5-37fdb9b04f8a> in <module>
      3 print("Generating mels")
      4 if generate_mels:
----> 5     create_mels()
      6 
      7 print("Checking for missing files")

3 frames
/content/tacotron2/stft.py in transform(self, input_data)
     82 
     83         # similar to librosa, reflect-pad the input
---> 84         input_data = input_data.view(num_batches, 1, num_samples)
     85         input_data = F.pad(
     86             input_data.unsqueeze(1),

RuntimeError: shape '[1, 1, 52781]' is invalid for input of size 10556```

it was touched by at least 4 separate people, 1 of whom disclaims responsibility for the code itself

all i know is something's very wrong with your audio

if you want, I can send you a print of the files to see how it turned out