Implement support for deepspeech v3 in ASR attack and Deepspeech estimator

[minhitbk]

Description

This PR implements support for Deepspeech V3 in ASR attack and the estimator. In addition, the Deepspeech V3 testing environment is also created.

Fixes #911
Fixes #1033

Type of change

Please check all relevant options.

• Improvement (non-breaking)
• Bug fix (non-breaking)
• New feature (non-breaking)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

Testing

Please describe the tests that you ran to verify your changes. Consider listing any relevant details of your test configuration.

• Test A
• Test B

Test Configuration:

• OS
• Python version
• ART version or commit number
• TensorFlow / Keras / PyTorch / MXNet version

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[codecov-commenter]

Codecov Report

Merging #1107 (10ed294) into dev_1.7.0 (2a402cd) will decrease coverage by 23.27%.
The diff coverage is 65.62%.

@@              Coverage Diff               @@
##           dev_1.7.0    #1107       +/-   ##
==============================================
- Coverage      64.45%   41.18%   -23.28%     
==============================================
  Files            221      221               
  Lines          19295    19328       +33     
  Branches        3533     3549       +16     
==============================================
- Hits           12437     7960     -4477     
- Misses          5438    10407     +4969     
+ Partials        1420      961      -459     

Impacted Files	Coverage Δ
...timators/speech_recognition/pytorch_deep_speech.py	72.75% <65.62%> (-1.08%)	⬇️
...rt/defences/detector/poison/clustering_analyzer.py	8.22% <0.00%> (-86.71%)	⬇️
...ces/detector/evasion/subsetscanning/scanningops.py	11.59% <0.00%> (-85.51%)	⬇️
art/attacks/evasion/carlini.py	8.40% <0.00%> (-80.87%)	⬇️
...ient_descent/projected_gradient_descent_pytorch.py	16.27% <0.00%> (-76.75%)	⬇️
art/metrics/verification_decisions_trees.py	13.68% <0.00%> (-73.16%)	⬇️
art/attacks/evasion/decision_tree_attack.py	22.66% <0.00%> (-69.34%)	⬇️
art/attacks/evasion/adversarial_patch/utils.py	11.53% <0.00%> (-69.24%)	⬇️
art/attacks/evasion/elastic_net.py	17.41% <0.00%> (-69.04%)	⬇️
art/defences/detector/poison/activation_defence.py	15.71% <0.00%> (-68.90%)	⬇️
... and 78 more

[beat-buesser]

Should we define a version for torch since we define an older version for torchaudio?

[minhitbk]

No, this is already defined with the base image.

[beat-buesser]

Is tensorflow required in this test image for PyTorch models?

[minhitbk]

Yes, for some "import" in the conftest.py file which is not created by this PR, without tensorflow, error occurs.

[beat-buesser]

Hi @minhitbk Thank you very much for adding support for DeepSpeech v3!

I have proposed a few small changes to account for recent CI updates and a few questions, what do you think?