Have you ever been in a store or restaurant and thought, I love this song, but what is it?â â there’s quite a few ways to find out. Mobile applications like SoundHound, Shazzam, and MusicID are driven by databases of audio waveforms from popular music. When you use the app and record a snippet of the song, the app draws a correlation between the waveforms in the database and the music happening around you to decipher which tunes you are rocking out to in the grocery store.
An important tool in signal measuring is tracking correlation. Using an oscilloscope and checking for correlation in a wavelength, you can parse out a signal from noise, compare frequency of two signals, measure delay between two signals â the tracking and subsequent application of correlation measurement is nearly endless. In this oscilloscope tutorial, we’re going to run you through the process of measuring correlation with an oscilloscope.
First we need a gauge by which to judge correlation of two signals. What we call the correlation coefficient is a score that rates how similar two signals are. For example if two signals have nothing in common, the correlation coefficient is 0. If the two signals are identical, the correlation coefficient is 1. If two signals are a mirror image of one another, the score is a -1.
Now, there are two types of correlation functions you can use an oscilloscope to track: Cross-Correlation and Auto-Correlation. The first being a comparison of two individual signals and the latter being a comparison of the same signal compared to phase shifted copies of itself.
Although correlation can be measured by hand and with a lot of math, there are easier tools now that may be native on your oscilloscope or available through a 3rd party.
Cross correlation is essentially a measure of time delay between two signals â this gives an assessment of the similarity between the two signals â the lower the time delays between each point, the more similar the signals are. This type of correlation testing is most often used to measure the time delay between two signals or to detect a short signal from a sample within a longer signal.
Because an auto-correlation measurement is over one signal, the analytics derived tell you primarily about pattern over the course of a signal. Auto-correlation can spot a pattern out of a sample waveform riddled with noise by picking up on underlying correlations in the noise that maybe you were not able to see.
Auto-correlation also measures periodicity â find out how often a signal repeats (or should repeat) itself over the course of a waveform and analyze the pattern of repetition within the signal.
Not all oscilloscopes are made equal. Some may have built in correlation measurement ability and some may need a logic analyzer, an additional piece of hardware with many applications including correlation measurement. Some oscilloscopes just need some 3rd party software to become correlation analyzers â or you could always break out your calculus book and figure out how to measure correlation by hand.