Aaaaaaa so that's how filtering and delay are related to each other. Damn, that really wasn't clear for me when I was reading up on it, these visualizations really help!
Include key and important concepts and explain in-depth in an easy-to-understand method, which can establish a strong framework and connection in the cognitive structure of learners. Very touching teaching, thank you very much! Excellent!!
Very clearly explained! Getting back to reviewing this stuff for some pet projects after 15 years. Hits different with clear explanations, nice diagrams, and a polished presentation. THANK YOU! :subscribed!:
Wow I've never thought about the LPF always obstructing fast changing signals in a way that it would smooth things out and become this "avg"-ing filter
The best part for me was the relationship of moving average to FIR, and running average to IIR, and how that affects frequency response. It sounds like you used two different microphones, and at one point I was confused as to whether the change in timbre was part of the lesson or, as I think was the case, just an artefact of the editing. The background marimba or whatever it was distracting when listening with headphones. I think the video would be better off without it. Overall, that's great content and I'm looking forward to watching more of them.
Fantastic video! Great explanation paired with great visuals, I actually understand how LP filters work now. Can't wait to mess around with the math behind this and hopefully make some interesting variations of the filter. Bit sad to see that this video is so low in views though. I think you could attract a lot more attention with a title like "How digital low-pass filters work", which is what I initially typed in to find information about how LP filters work.
I tried manim initially, but it didn't really fit with my process and experience (largely based in matplotlib). So for the graphs, I generate the individual frames with matplotlib, compile those into short animation videos, and put the whole thing together with narration in macOS Keynote. I probably spend way too much time tweaking individual animation frames 🙂 The whole process takes a while, but I like how it looks in the end!
Posted No. 7 a while ago and I *just* posted No. 5. Hope you like them 🙂 I'm teaching the course again this Fall, so there will be more updates coming soon!
Found a typo I think. On 10:57, since h(n) = (0.9)^n, then its H(z) = 1/(1-0.9/z). Yet, applying the ZT to the provided difference equation: y(n) = 0.9x(n)-y(n-1), we get H(z) = 0.9/(1-0.9/z).
At 12:15, I think the two graph lines should be getting further apart, not coming closer together! The magnitude of higher frequencies should be reducing as the filter graph gets steeper, not the other way round. Actually, to follow your narration, the filter graph should start of steep and then get flatter and flatter (closer to 1).
I believe the animation is correct, but I probably could've done a better job explaining this... As the coefficient changes (decreases), the time-domain impulse response (top graph) becomes closer to an impulse, but the frequency response (bottom graph) becomes closer to an all-pass filter (flat), which is what you described. Hope that makes sense!
Love these vids, its helping me revise my signals and systems and DSP. At 10:46 shouldn’t it be n+1-m as at m=n ill get 0.9 otherwise i would get 1*x[n] at n=m.
I wish I had seen this video earlier
taught me dsp more than the teachers did in the entire semester
Got up at 6 am to work on an assignment I have been stressed with and the music made me laugh in 15 sec... Thanks! I NEED this
I feel glad to find this channel, tks for teach us
Dude this is so well made
Thank you loads for the uploads
A superb video! We need more videos like these! Looking forward to the rest of the series.
Aaaaaaa so that's how filtering and delay are related to each other. Damn, that really wasn't clear for me when I was reading up on it, these visualizations really help!
That was really well explained, loved the visuals!
I've been trying to understand this for a whole semester, and gave up somewhere in the middle. You're really good at this.
It's amazing this could be done in 1987 with musical instruments like the Roland D50.
Include key and important concepts and explain in-depth in an easy-to-understand method, which can establish a strong framework and connection in the cognitive structure of learners.
Very touching teaching, thank you very much! Excellent!!
WOW THIS IS SO CLEAR
good stuff Dr. Kim!
I love you bro. I've been struggling to understand this for weeks!
After Long time saw a nice tutorial. Thank u sir
I am speechless, what quality content!
Wow that was one of the best explanations and visualization of convolution I've seen. Thank you!
This, is AMAZING, thank you!
This is such an excellent clear explanation of FIR and IIR filters! Good job.
Very clearly explained! Getting back to reviewing this stuff for some pet projects after 15 years. Hits different with clear explanations, nice diagrams, and a polished presentation. THANK YOU! :subscribed!:
Awesome visualizations, explanations
Fantastic series! Thank you very much for sharing these.
This song from 5:15 is something from Daredevil ,am i right ?
i'm sure i heard it multiple times when fisk is around .
Probably (sorry, I haven't seen Daredevil). It's Bach's Cello Suite No. 1, as performed by Yo Yo Ma.
Such a good series! Upvote for the continuation of this amazing series/course!
Wow I've never thought about the LPF always obstructing fast changing signals in a way that it would smooth things out and become this "avg"-ing filter
Great explanation, thanks!
Now almost one year after my signal class in college I finally can visualize convolution
pure gold
The best part for me was the relationship of moving average to FIR, and running average to IIR, and how that affects frequency response. It sounds like you used two different microphones, and at one point I was confused as to whether the change in timbre was part of the lesson or, as I think was the case, just an artefact of the editing. The background marimba or whatever it was distracting when listening with headphones. I think the video would be better off without it. Overall, that's great content and I'm looking forward to watching more of them.
Fantastic video! Great explanation paired with great visuals, I actually understand how LP filters work now. Can't wait to mess around with the math behind this and hopefully make some interesting variations of the filter.
Bit sad to see that this video is so low in views though. I think you could attract a lot more attention with a title like "How digital low-pass filters work", which is what I initially typed in to find information about how LP filters work.
Great video!
I found this video very usefull.
Thank you for a great vid. Very clear and well explained, great animations too
Захоплююча лекція. Переглянув з задоволенням.
These are extremely well edited videos with excellent explanations!
This is amazing!
great！
Perfect 👏🏻👏🏻👏🏻!
Can you also make a video of low pass filters and how to eliminate or avoid overshoots in the time domain?
Amazing series. Why is the convolution sum from negative to positive infinity?
Superb vdo
Dear Youngmoo Kim,
I was wondering how was the animation made? It is so cool. Was it made by manim?
I tried manim initially, but it didn't really fit with my process and experience (largely based in matplotlib). So for the graphs, I generate the individual frames with matplotlib, compile those into short animation videos, and put the whole thing together with narration in macOS Keynote. I probably spend way too much time tweaking individual animation frames 🙂 The whole process takes a while, but I like how it looks in the end!
I wonder what happened to Applied DSP No. 5 and 7
Posted No. 7 a while ago and I *just* posted No. 5. Hope you like them 🙂 I'm teaching the course again this Fall, so there will be more updates coming soon!
Found a typo I think. On 10:57, since h(n) = (0.9)^n, then its H(z) = 1/(1-0.9/z). Yet, applying the ZT to the provided difference equation: y(n) = 0.9x(n)-y(n-1), we get H(z) = 0.9/(1-0.9/z).
At 12:15, I think the two graph lines should be getting further apart, not coming closer together! The magnitude of higher frequencies should be reducing as the filter graph gets steeper, not the other way round. Actually, to follow your narration, the filter graph should start of steep and then get flatter and flatter (closer to 1).
I believe the animation is correct, but I probably could've done a better job explaining this... As the coefficient changes (decreases), the time-domain impulse response (top graph) becomes closer to an impulse, but the frequency response (bottom graph) becomes closer to an all-pass filter (flat), which is what you described. Hope that makes sense!
@@youngmoo-kim Yes, sorry, I'm now not sure why I was confused/wrong about this 6 months ago.
What if I have to filter my data offline? Is it work if I apply FFT, cut off the frequencies I don't need, then I apply inverse FFT?
You can, but there are some hidden issues. See video No. 8 for more details!
Love these vids, its helping me revise my signals and systems and DSP. At 10:46 shouldn’t it be n+1-m as at m=n ill get 0.9 otherwise i would get 1*x[n] at n=m.
what is the bgm at the beginnig? i love it
Where did Applied DSP No. 5 go?
Just posted it 🙂 Sorry for the long wait. I'm teaching the course again this Fall, so there will be more frequent updates in the near future!
Wow, I just have question, how IIR filter can over come aliasing error?
The key to reducing aliasing is low-pass filtering. So you can use either an FIR or IIR to design a low-pass filter to do the job.
I would love to learn this code
It would be better if you add subtitle.
Pls add subtitle