ibm / max-audio-sample-generator Goto Github PK

the README has the following link to where the models files are stored (IBM Cloud Object Storage):

• http://max-assets.s3-api.us-geo.objectstorage.softlayer.net/max-audio-sample-generator/wavegan.tar.gz

but that link fails! alternatively, the Dockerfile has the link:

• http://max-assets.s3-api.us-geo.objectstorage.softlayer.net/max-audio-sample-generator/models.tar.gz

which works.

This line write a wave file to a constant file name:

This may cause race condition and is not so efficient.

assert audio_range > 0.05 should be set less strict

Recent failures:

>           assert audio_range > 0.05  # this is a very small range, the model should be improved in the future
E           assert 0.018829584 > 0.05

>           assert audio_range > 0.05  # this is a very small range, the model should be improved in the future
E           assert 0.029271167 > 0.05

>           assert audio_range > 0.05  # this is a very small range, the model should be improved in the future
E           assert 0.036722984 > 0.05

>           assert audio_range > 0.05  # this is a very small range, the model should be improved in the future
E           assert 0.037834674 > 0.05

>           assert audio_range > 0.05  # this is a very small range, the model should be improved in the future
E           assert 0.0016102523 > 0.05

README.md contains these lines:

$ curl -X GET 'http://localhost:5000/model/predict' -H 'accept: audio/wav' > result.wav && open result.wav

$ curl -X GET 'http://localhost:5000/model/predict?model=stop' -H 'accept: audio/wav' > stop.wav && open stop.wav

however open is a command specific to OSX, we should remove this or specify it only works on Mac

The play button seems to work in Google Chrome but not Firefox

Verify the MD5 hashes of files downloaded from COS are correct