[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

2022-05-12: Ghost posting is now globally disabled. 2022: Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/diy/ - Do-It-Yourself


View post   
View page     

[ Toggle deleted replies ]
File: 537 KB, 1600x1200, 1638993684143.jpg [View same] [iqdb] [saucenao] [google]
2470918 No.2470918 [Reply] [Original] [archived.moe]

Would a fourier transform or wavelet transform be better for characterizing the sound of tapping on different surfaces? I'm trying to train a machine learning model to identify different tapping sounds. So far I've used a canned FFT function on raw audio data and have been able to train one to tell a desk tap from a keyboard key press almost every time. Although taps on similar sounding surfaces have proven more difficult. Would wavelets be better? I know that Fourier transforms are best for stationary signals whose frequencies don't change over time. I can't imagine these sounds have unwavering frequencies, but perhaps they are brief enough to where that doesn't matter?

>> No.2470989

>>2470918
a tap is more than 1 frequency, usually a delicious combo with a distinct time to decay for particular frequencies.
>wavelets
id assume they wouldn't change frequencies but youre being autistically vague. u tapping a table? cymbal? some ass?
My approach would be looking for particular fft values by freq, amp, phase and rate of decay. wavelets are possible but sound like alot more work

>> No.2472558

>>2470918
I recommend Dynamic Time Warping (DTW).

>>
Delete posts
Password [?]Password used for file deletion.
Captcha
Action