Fix Deprecation/Future Warnings in Notebook 211-Speech-to-Text

[YDX-2147483647]

• waveplot is deprecated in favor of waveshow since librosa v0.8.1, and removed in v0.9.0.

  This was mentioned in the original pull request in Dec. 2021, but not fixed (for unknown reason).

  Note that max_sr in waveplot is removed in waveshow. max_sr controls how signals are down-sampled if they’re too long. (waveshow uses “a downsampled amplitude envelope” instead)

In the committed version, Imports are at the top of the notebook.

• librosa.filters.mel in audio_to_mel

  FutureWarning: Pass sr=16000, n_fft=512 as keyword args. From version 0.10 passing these as positional arguments will result in an error.

Question

The original pull requeset did NOT add librosa (the audio analysis package used here) into requirements.txt or .docker/Pipfile. Is it on purpose? Should I tell how to install it in the notebook?

[YDX-2147483647]

Sorry, it's mine. max_points in waveshow should be an int, not float.

(I checked it on a private repo and manually copy the changes to this repo - the screenshot was right.)

[Debskij]

Hi @YDX-2147483647, thanks for your contribution. It would be great if you can add librosa to both requirements.txt and Pipfile (and lock it afterward).

[YDX-2147483647]

New discovery: librosa v0.8.1 exists in Pipfile.lock.

I pip install again and found it required by ppgan v2.1.0.

> pip show librosa
Name: librosa
Version: 0.8.1
……
Requires: audioread, decorator, joblib, numba, numpy, packag
ing, pooch, resampy, scikit-learn, scipy, soundfile
Required-by: ppgan

openvino_notebooks/requirements.txt

Line 26 in 6e76762

ppgan==2.1.*

ppgan's requirements.txt:

librosa==0.8.1

---

It seems that ppgan only uses librosa to read/write wav. Therefore I'll turn to ppgan and mark this PR as WIP.

[adrianboguszewski]

@YDX-2147483647 please add it to requirements and Pipfile anyway

[YDX-2147483647]

✓ OK, but maybe tomorrow.

[Debskij]

Hi @YDX-2147483647. Thanks for updating requirements.txt and Pipfile, still I can see that you haven't updated the Pipfile.lock file. To do that you have to (on your own machine) get into the directory openvino_notebooks/.docker and run command pipenv lock. Please add the updated file to the repository when it is ready.

[YDX-2147483647]

Uhh… Thanks for helping me. In fact I had noticed “(and lock it afterward)”, but got stuck at pipenv install for hours (even on main). I am working on it.

[YDX-2147483647]

Could I leave Pipfile.lock as it is?

Reason: As mentioned above, there’s no new requirement at all — we only elevate librosa from ppgan's sub-dependency to a direct dependency. Current Pipfile.lock does contain librosa (v0.8.1), making it a valid environment.

My difficulty: I can’t pipenv install even on main because of PyTorch.

My tries

(Failures omitted)

• HTTP Errors.

  I use a mirror source instead, then manually change Pipfiles back before I commit.

• PyPI can't recognize +cpu for torch and torchvision.

  With everything else settled, I pipenv run pip install from PyTorch's site.

• pipenv lock failed. (pipenv lock without PyTorch succeeded.)

  Use advanced syntax in Pipfile.

• No space left on device.

  Clean pip cache. Used up again. Clean. Again. …

  (Maybe a pipenv issue: Pipenv downloads PyTorch for all versions of Python, grabbing 16GB of data instead of just 1.7GB)

It sounds ridiculous that I am fighting against PyTorch environment days and night without using it… But it's truth. ╮(╯_╰)╭

[adrianboguszewski]

@YDX-2147483647 ok, we will add librosa in another PR, let's merge your changes in 211

[YDX-2147483647]

Thanks!