fix(data_utils): allow float32 audio to be processed properly

[BlueAmulet]

In an attempt to cut down on noise introduced by int16 wav quantization, I wanted to try using float32 wav files instead.

I found that python's wave library does not understand float32, so I removed the wave library based duration code for librosa's get_duration function instead.

I was also getting a WavFileWarning: Chunk (non-data) not understood, skipping it. message from scipy.io.wavfile.read. I noticed that the load_wav_to_torch() function + the normalization and unsqueeze, results in the exact same format as a simple torchaudio.load(), so this code has been simplified as well.

[codecov-commenter]

Codecov Report

Merging #115 (e597a9a) into main (baf58d2) will decrease coverage by 0.14%.
The diff coverage is 66.66%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

@@            Coverage Diff             @@
##             main     #115      +/-   ##
==========================================
- Coverage   16.80%   16.67%   -0.14%     
==========================================
  Files          28       28              
  Lines        3391     3382       -9     
  Branches      393      391       -2     
==========================================
- Hits          570      564       -6     
+ Misses       2810     2807       -3     
  Partials       11       11              

Impacted Files	Coverage Δ
src/so_vits_svc_fork/preprocess_flist_config.py	89.39% <50.00%> (-1.02%)	⬇️
src/so_vits_svc_fork/data_utils.py	17.97% <66.66%> (+1.49%)	⬆️
src/so_vits_svc_fork/__init__.py	100.00% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[34j]

@allcontributors add BlueAmulet code

[allcontributors]

@34j

I've put up a pull request to add @BlueAmulet! 🎉