search

jcvasquezc / DisVoice

feature extraction from speech signals
https://disvoice.readthedocs.io/en/latest/
MIT License
354 stars 80 forks source link

ValueError: zero-size array to reduction operation minimum which has no identity #39

Open Cybernorse opened 2 years ago

Cybernorse commented 2 years ago

I have been trying to extract glottal features from cough sounds but i am unable to do so, i get the error titled above, i want to know if there is any way possible to extract glottal features from cough sounds, Thank you

mcvarer commented 2 years ago

up up up

mcvarer commented 2 years ago

Have you fixed @Cybernorse

Cybernorse commented 2 years ago

Hi there ! mcvarer, Lowering the sampling rate makes the code work, but it completely changes the sound itself, as far as i remember the debugging, it had problem extracting formant F0 from some of the sound signals, so when it tries to find the minimum value in the output array of F0 formant this error is thrown because the array is empty where it shouldn't be, i think the formant features model is based on something that only extracts speech vocalics, maybe source filter model, i did read somewhere that lowering the sampling rates does reveal vocal voices, so it makes sense that it starts working on some of them it was not working on before, i was actually able to extract other vocal based features like LPC from the cough sound so i had to ignore disvoice as our project progressed, that's all i have to say.