Convolution in the time domain

Now that you understand the Fourier transform, it's time to start learning about time-frequency analyses. Convolution is one of the best ways to extract time-frequency dynamics from a time series. Convolution can be conceptualized and implemented in the time domain or in the frequency domain. It is important to understand both conceptualizations. We start with the time domain implementation.

[yijiang]

How can anyone dislike this video? After days of searching this is the best series I found so far!!!

[lazygirlrants]

I am doing 3rd course by Mike on Udemy and it's brilliant. We love you, Mike!

[Ameen_AAA]

Thank u for the clarification. YouTube experts r saving my life again 😅. For those who r in a hurry, u can play it at 2x.

[giselecamargo]

The best explanation I've ever seen on the subject!

[xiangzhang]

3:45 This is the best way to understand why wavelet can extract "time-frequency" features!

[dewimahzyafortuna]

THW BEST EXPLANATION I'VE EVER SEEN IN MY LIFE..

[andreacervo]

Hi, first of all thanks and compliments for the videos.
I am a bit confused to join what I learned in your videos and the theory of the wavelet transform. I am reading on different books. how is the convolution analysis related to the wavelet transform?
It seems that convolution implicitly slides the wavelet along the signal. So what’s the sliding factor m defined in the wavelet formula and is there any reason why you never include it in your examples?
Thanks and compliments again

[ianh]

Thanks for this I was just talking about this with a colleague.

[wjz]

Hi Mike, great course. I have question for you:

Computing wavelet in point N of source data reqiues to compute convolution betwen wavelet (Morlet for example) and samples from N-x to N+x, where 2x+1 is discrete wavelet length. In many cases, eg. in market trend analysys, we don't have future samples. In your opinion is posible to create asymetrical wavelet whose we can convolute whith samples N-x to N? Is Haar wavelet is that what i need for zero-latency wavelet analysys?

[christiansetzkorn]

Thanks very nice graphical explanation!

[igorcruz]

Hi Mike. Your collections of videos are awesome! It led me to learn so much in a considerable narrow period of time. Thanks a lot! However, I am analysing data with Wavelet Leaders and no coefficients (i.e, DWT Leaders for Multifractal analysis). Is it possible to adapt with a bit of code your script to get the "leaders"? Tks

[chaoh]

Hi Mike, thanks for the videos! I think there is an error in line 44 of the script, where you have:

I think it should be:
ti=1:length(dat4conv) - (length(kernel) - 1)
although in this case, they don't make a difference.

[sam-zydn]

Thanks Mike. Why do you say convolution? This is just correlation however, if the wave let does not symmetric still you are using correlation.

[Khwartz]

Thanks for your Clear and Sound Video :)

I have a rather personal question to ask You: have You ever been Actor in the Star Trek NG Series? My hears find You voice Most Similar to the one of an Actor who had plaid a character named "The Traveller"; I had this episode, so that would be Really Fun that would have been you! ;-)