Example for using Wav2Vec2 SpeechToText from Huggingface

[altre]

Description

This is a simple step by step example of how to use torchserve to download and serve a speech to text model from huggingface.

Fixes #1656

Type of change

Added an example

Feature/Issue validation/testing

Output of following the instruction steps should be as following:

❯ conda env create -f environment.yml
Collecting package metadata (repodata.json): done
Solving environment: done
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
Installing pip dependencies: \ Ran pip subprocess with arguments:
['/Users/alanschelten/miniconda3/envs/wav2vec2env/bin/python', '-m', 'pip', 'install', '-U', '-r', '/Users/alanschelten/serve/examples/speech2text_wav2vec2/condaenv.r8y4m9_t.requirements.txt']
Pip subprocess output:
Collecting torch-workflow-archiver
  Using cached torch_workflow_archiver-0.2.4-1-py2.py3-none-any.whl (12 kB)
Collecting torch-model-archiver
  Using cached torch_model_archiver-0.6.0-py2.py3-none-any.whl (14 kB)
Collecting torchserve
  Using cached torchserve-0.6.0-1-py2.py3-none-any.whl (19.6 MB)
Collecting enum-compat
  Using cached enum_compat-0.0.3-py3-none-any.whl (1.3 kB)
Collecting future
  Using cached future-0.18.2-py3-none-any.whl
Collecting psutil
  Using cached psutil-5.9.3-cp310-cp310-macosx_11_0_arm64.whl (243 kB)
Requirement already satisfied: wheel in /Users/alanschelten/miniconda3/envs/wav2vec2env/lib/python3.10/site-packages (from torchserve->-r /Users/alanschelten/serve/examples/speech2text_wav2vec2/condaenv.r8y4m9_t.requirements.txt (line 3)) (0.37.1)
Requirement already satisfied: packaging in /Users/alanschelten/miniconda3/envs/wav2vec2env/lib/python3.10/site-packages (from torchserve->-r /Users/alanschelten/serve/examples/speech2text_wav2vec2/condaenv.r8y4m9_t.requirements.txt (line 3)) (21.3)
Collecting Pillow
  Using cached Pillow-9.3.0-cp310-cp310-macosx_11_0_arm64.whl (2.9 MB)
Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /Users/alanschelten/miniconda3/envs/wav2vec2env/lib/python3.10/site-packages (from packaging->torchserve->-r /Users/alanschelten/serve/examples/speech2text_wav2vec2/condaenv.r8y4m9_t.requirements.txt (line 3)) (3.0.9)
Installing collected packages: torch-workflow-archiver, enum-compat, psutil, Pillow, future, torchserve, torch-model-archiver
Successfully installed Pillow-9.3.0 enum-compat-0.0.3 future-0.18.2 psutil-5.9.3 torch-model-archiver-0.6.0 torch-workflow-archiver-0.2.4 torchserve-0.6.0

done
#
# To activate this environment, use
#
#     $ conda activate wav2vec2env
#
# To deactivate an active environment, use
#
#     $ conda deactivate

Retrieving notices: ...working... done
❯ conda activate wav2vec2env
❯ ./download_wav2vec2.py
Some weights of Wav2Vec2ForCTC were not initialized from the model checkpoint at facebook/wav2vec2-base-960h and are newly initialized: ['wav2vec2.masked_spec_embed']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
❯ ./archive_model.sh
❯ torchserve --start --model-store model_store --models Wav2Vec2=Wav2Vec2.mar --ncs
TorchServe is already running, please use torchserve --stop to stop TorchServe.
❯ torchserve --start --model-store model_store --m
❯ torchserve --stop
TorchServe has stopped.
❯ torchserve --start --model-store model_store --models Wav2Vec2=Wav2Vec2.mar --ncs
<...server output...>
❯ curl -X POST http://127.0.0.1:8080/predictions/Wav2Vec2 --data-binary '@./sample.wav' -H "Content-Type: audio/basic"
I HAD THAT CURIOSITY BESIDE ME AT THIS MOMENT% 

Checklist:

• Did you have fun?
• Have you added tests that prove your fix is effective or that this feature works?
• Has code been commented, particularly in hard-to-understand areas?
• Have you made corresponding changes to the documentation?

[msaroufim]

Very nicely done, thank you!

[agunapal]

Thanks @altre for the PR . Looks good overall. Please include the output of the sample example in the README

[altre]

Ok, @agunapal , added the output

[agunapal]

looks good. Thanks @altre

[codecov]

Codecov Report

Merging #1939 (b6edf9e) into master (33e1e97) will not change coverage.
The diff coverage is n/a.

❗ Current head b6edf9e differs from pull request most recent head 8f72ac0. Consider uploading reports for the commit 8f72ac0 to get more accurate results

@@           Coverage Diff           @@
##           master    #1939   +/-   ##
=======================================
  Coverage   44.95%   44.95%           
=======================================
  Files          63       63           
  Lines        2609     2609           
  Branches       56       56           
=======================================
  Hits         1173     1173           
  Misses       1436     1436           

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