07-22-2016 09:19 AM
Hello everyone,
I think my subject is redundant, but my researches did not satistfied me to have a precise answer.
I want to compare a vocal signature. And for it, I am going to use a FFT which will compare the saved and the analyzed spectrum.
For now, I managed to record the sound and then I can the signal. I am little short of ideas concerning the comparison of the signals.
You will find my VI inclosed.
Thank you for being willing to support me to end this VI.
07-22-2016 09:30 AM
When analyzing speech or other time-varying sounds, you need to ask how the spectrum changes with time. This has been pretty widely studied, and is a little more complicated than just taking a (single) FFT. Look up "spectrogram" ...
Bob Schor
07-22-2016 10:48 AM
Spectrogram is what you're looking for. It basically takes the time domain signal and splits it up into (sometimes) overlapping "bins" of which each the FFT is taken. The result may look somewhat familiar:
It's the way your car stereo/av receiver/music program makes those bars that go up and down with the beat.
08-01-2016 09:57 AM
Ok thank you ! I kown what you mean. Does someboby have exemples of how i can do with the spectogram ?
08-05-2016 07:15 AM
Hello,
I think I am on the right way now. I can observe a stable spectrogram for the same word, like for example " Yes or No".
The last step, is to compare a saved spectrogram with a just acquired one. Do you have any idea how I can do please?
You'll find attached my VI.
PS: Sometimes I have a blank during the record and an error appear. Do you have any idea how I can make my VI better?