how to test the network on new voices?

[ranch-hands]

Hi
thanks for your good work .
will you please explain how to test the network on new data ?
how can we test it on our own voices?

thanks

[SuperKogito]

Hello,
The easiest way is to use https://github.com/SuperKogito/Voice-based-gender-recognition/blob/cecc0469196e0a212716a1b3ddc085fe8e78288a/Code/GenderIdentifier.py#L66C1-L68C32

if __name__== "__main__":
    gender_identifier = GenderIdentifier("TestingData/females", "TestingData/males", "females.gmm", "males.gmm")
    gender_identifier.process()

just place your data under a TestingData folder and pass it to GenderIdentifier() as shown above.

[ranch-hands]

Thanks a lot .

…

On Wed, Jul 5, 2023 at 12:06 PM Ayoub Malek ***@***.***> wrote: Hello, The easiest way is to use https://github.com/SuperKogito/Voice-based-gender-recognition/blob/cecc0469196e0a212716a1b3ddc085fe8e78288a/Code/GenderIdentifier.py#L66C1-L68C32 <http://Code/GenderIdentifier.py> if __name__== "__main__": gender_identifier = GenderIdentifier("TestingData/females", "TestingData/males", "females.gmm", "males.gmm") gender_identifier.process() just place your data under a TestingData folder and pass it to GenderIdentifier() as shown above. — Reply to this email directly, view it on GitHub <#13 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASMJBK3JI5EZT4XRG25BFRTXOURRTANCNFSM6AAAAAAZ6TJCWI> . You are receiving this because you authored the thread.Message ID: ***@***.*** com>

[ranch-hands]

It means that each time (for testing) the network should be trained?
each time the run.py downloads the dataset and builds the TestingData folder by itself.

[SuperKogito]

No you train once and then you test as many times as you want.
the https://github.com/SuperKogito/Voice-based-gender-recognition/blob/master/nnCode/ModelsTrainer.py is responsible for the training, which results in having 2 models one for males and one for females. These models are used by the https://github.com/SuperKogito/Voice-based-gender-recognition/blob/master/nnCode/GenderIdentifier.py to test the new voices/ audios.

[ranch-hands]

When I give a sound to GenderIdentifier.py it gives error:

984/984 [==============================] - 2s 2ms/step - loss: 0.2400 - accuracy: 0.9044
Epoch 5/5
984/984 [==============================] - 2s 2ms/step - loss: 0.2262 - accuracy: 0.9094

--> TESTING        : testsound.m4a
Error

--> TESTING        : all_systems_go.wav

Some rows of transmat_ have zero sum because no transition from the state was ever observed.
Some rows of transmat_ have zero sum because no transition from the state was ever observed.
Some rows of transmat_ have zero sum because no transition from the state was ever observed.
Some rows of transmat_ have zero sum because no transition from the state was ever observed.
(16, 39)
Error
*** Accuracy = 100.0% ***

or:

Epoch 5/5
984/984 [==============================] - 2s 2ms/step - loss: 0.2195 - accuracy: 0.9144
--> TESTING        : testsound.wav
Error
--> TESTING        : all_systems_go.wav
(16, 39)
Error
*** Accuracy = 100.0% ***

[SuperKogito]

Your audios must be mono waves as far I remember so anything different will probably not work. Make sure your audio is not stereo. Use ffmpeg to convert it if needed.

[alizahid999]

thanks dear, connect you very soon.

…

On Tue, Jul 18, 2023 at 9:08 PM Ayoub Malek ***@***.***> wrote: Your audios must be mono waves as far I remember so anything different will probably not work. Make sure your audio is not stereo. Use ffmpeg to convert it if needed. — Reply to this email directly, view it on GitHub <#13 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AUEXXD36FP76HGTEV7HI5ODXQ2YIZANCNFSM6AAAAAAZ6TJCWI> . You are receiving this because you are subscribed to this thread.Message ID: <SuperKogito/Voice-based-gender-recognition/issues/13/1640525049@ github.com>