Cheat Engine

HomerSexual

Hello all:

I've been working on a project for a little while now and one of the components of the project requires determining the rhythm of sound input. The way that I plan to go about this, until I determine a better way, is to take the time in between each successive note. However, sound is a wave and goes up and down, and so it is not easy to find the relative peaks.

Screenshot 1:

Uploaded with ImageShack.us

As you can see, the global peaks are very clear to the eye, but I can't find an algorithm to determine them.

Help/ideas?

Varreon · Posted: Thu Jun 09, 2011 9:29 pm Post subject:

Look into the fourier transform.

HomerSexual · Posted: Thu Jun 09, 2011 9:30 pm Post subject:

A Fourier Transformation will determine the frequency of a sound interval, however, it will not pick out the peak from a set of data.

I plan to use FFT's to determine frequency, but rhythm must be determined using the amplitude of the wave and looking for rapid spikes.

Slugsnack · Posted: Fri Jun 10, 2011 2:52 am Post subject:

smooth it by a certain factor ? eg. cut out the granularity by half or something

b6oy · Newbie cheater Reputation: 0 Joined: 21 Sep 2009 Posts: 10

Simplest way, if your goal is rhythm recognition. Humoments with Bayes classifier and alot of samples.

HomerSexual · Posted: Fri Jun 10, 2011 10:32 am Post subject:

I do not want to use a learning algorithm for determining the rhythm. The way I want to do it for now is to get the time where every new note is plucked (using a bass track so plucked works here) and to determine the time between successive notes.

How would I go about smoothing the data to remove granularity? I can square the data in order to remove negatives while still preserving peaks, but the data is still up/down at that point, but it removes the problem of pos/neg successive data.

Edit:

Here is the data with a square filter

Uploaded with ImageShack.us

AhMunRa · Posted: Fri Jun 10, 2011 2:46 pm Post subject:

This may give you some ideas, or at least in the right direction.

http://stackoverflow.com/questions/79445/beats-per-minute-from-real-time-audio-input

Of course this is only good if all you want is bpm, if I understand you want the actual rhythm though.

HomerSexual · Posted: Fri Jun 10, 2011 6:17 pm Post subject:

thanks, that's a good starting post! It's not quite what I am looking for, but it's something close that I can probably adapt