Server Error

[rahulkarda]

Server error. Either the daily quota of the server is over or there was some internal error. You can try to generate a new sample in a few seconds. If the error persist, try comming back tomorrow or download the local version from Github :)

[userMak]

are you trying to play it locally? if not have you found a solution?

[rahulkarda]

are you trying to play it locally? if not have you found a solution?

Yes running it locally. No solution found yet.

[userMak]

that's a same. We are so close and so far at the same time

[pratapyash]

@Thiagohgl, Is there any solution to this? I've been attempting to run it locally, but I consistently encounter server error 500.

[nguyenchiminhhieu]

@Thiagohgl I also encountered the same problem. Please give me a solution. Thank you

[Alehandreus]

@Thiagohgl I also encountered the same problem. Please give me a solution. Thank you

I did some digging, added try - except and located the error.
It occurs at nn_output = self.model(audio) line in AIModels.py file:

RuntimeError: stft requires the return_complex parameter be given for real inputs, and will further require that return_complex=True in a future PyTorch release.

Looks like silero models do have problems with newer PyTorch versions.

I got it working on torch 1.13.1.

[nguyenchiminhhieu]

@Alehandreus

Thank you for your enthusiasm. I have downgraded the version of pytorch to 1.13.1. Run on localhost after recording is finished, it displays as a picture ...

[Alehandreus]

There is a function GetAccuracyFromRecordedAudio in webApp.py that is responsible for sending the audio.
You can add try-except here and see what it prints (just like I did):

@app.route(rootPath+'/GetAccuracyFromRecordedAudio', methods=['POST'])
def GetAccuracyFromRecordedAudio():
    try:
        event = {'body': json.dumps(request.get_json(force=True))}
        lambda_correct_output = lambdaSpeechToScore.lambda_handler(event, [])
    except Exception as e:
        import traceback
        print(e)
        print(traceback.format_exc())
    return lambda_correct_output

Chances are this time it's about np.int which is deprecated. You need to replace two occurrences of np.int with int in WordMatching.py.

[nguyenchiminhhieu]

@Alehandreus It's awesome. Can you give me a reference to this project of yours

[Alehandreus]

Somebody have already fixed np.int issue: fork, pull request
If you need the exact place to insert try-except, you can find it in my fork or in commit changes