The bit at the end showing the coefficient and tap relationship is the highlight of the video :-). For me it tied this whole thing together! Thanks again for the brilliant videos.
Ah! I'm glad that solidified it for you! Seeing the impulse response as the coefficients is a key way for understanding FIR filters I suppose. You're welcome, and thanks for checking it out!
At 11:50 the flow chart shows z-1 for each tap. Does each tap operate on the same sample? I seem to remember it might need to operate on older samples, so z-1, z-2, z-3 and so on.
@@legosteveb a signal which flows through 3 z-1 delay blocks is equivalent to it passing through a singal z-3 delay block. As the signal passes through the circuit, each tap adds a sample of delay, and the same signal is sent to the next tap.
I don't have much about this knowledge background but your videos explanation are so intuitive to understand, helps a lot, hope to see your next video soon!
This is exactly what I have been looking for. I got into audio programming recently and I have struggled do find good resources that are understandable at my level of math knowledge. Amazing. Thank you!
Hi! Thanks for this beautiful set of video! @Akash Murthy, regarding your comment @08:50, a partial answer I can give you is from what I remember of my courses of electronics, especially automation, at school. Most of the time, in analog filters, the transfer function appears to follow the rule that the phase looks like the derivative of the gain. I said most of the time, but I do not guarantee this always applies, besides, sometimes you may have an offset of the phase. Translated, if you have a flat gain response, whether it is 0dB or another value, you should have a 0deg phase dB. If you have a constant -6dB/oct or -20dB/dec slope, you should have a constant (flat) -90deg phase. With a constant -12dB/oct or -40dB/dec you should have a constant (flat) -180deg phase, etc... At the cutoff frequency of a first order, which should be ~3dB, you should have a phase of -45deg. And so on... I can't say whether it applies or not in digital filters, most of all, FIR filters but it may be interesting to find out. @08:50, it appears that this theory applies.
Sorry. What I talked about earlier was essentially without active/discrete components... if you have resistors and capacitors and inductors, it should still be true. However, when you add an op amp, it becomes the higher the frequency (and closer to the 0dB open loop gain), the less true this statement is. I guess that in digital signal it should be the same somehow... the closer to Shannon frequency, the more non linear is the phase even if your gain is flat all the way... I mean, if you have a sampling frequency of 384kHz and handle audio frequencies up to 12kHz, you should be quite fine. Do you know of free Softwares that I can use to test digital filters? Do you know how yo calculate coefficients of the transfer function in Z of digital IIR filters like Butterworth, Bessel, Chebychev filters with input parameters such as order of filter, cutoff freq, sampling freq, etc? Thanks.
kudos for making such fine explainer series covering the fundamentals, it helps build intuition and serve as a great introduction to the topic ahead, hope you continue with this work
Ah, you're the first to bring up colour coding, thank you for fixating on that! When UA-cam throws random video recommendations down your way, it just makes sense to order similar content by colour.
So it shifts phase to cancel out the signal, and that how its filtered. I would call this Phase Filtering. The delay is relative to the sample rate, filtering the signal. Thank you so much, best tuts ever.
Thanks. Why education system doesn't have professors like you? Thanks a lot brother to you and your team for such wonderful content. Waiting for IIR filter.
What an awesome video. I absolutely love the conclusion. One of my biggest "aha!" effects ive had in the topic of filters. Also very nice graphics and demonstrations. The FIR Filter plugin part of the viodeo could´ve used a small overview in which all the values are presented at once at the end. For example the combination +0.5 and +0.5, -0.5 and +0.5, -0.5 and -0.5 and so on in one overview. I understood every word, even though im a german. Very good work and a big thank you.
That's awesome to hear, thanks very much for the feedback. Yes, I think I could've spent a little more time showing the plugin , with different values.
These videos are great, ive really enjoyed watching them. The production is amazing. How do you do all your animation, camera movement etc? Its really slick.
Admittedly I haven't finished your series of videos and may need to re-watch a few already, but am I correct in thinking feed foward filters are non-resonant? I'm a little curious how resonance works with convolution, but maybe that's explained in another video. Excellent videos, BTW. Your efforts here are greatly appreciated.
Congrats! very good video, animations, presentation, programing and examples. Can I ask you what app or program did use for presentation? Thanks for the video!
Thanks very much! I don't know if I want to do audio codec specific topics, it doesn't interest me very much unfortunately. I'm more interested in general DSP topics.
@@akashmurthy I am really looking forward to the rest videos of this topic. This is the first time I can understand a little bit concept on DSP. Your videos are super helpful for me. Thank you very much!
Not yet, I've been a bit busy over the last year. Haven't got around to it. But I'm making it right now, so it should be out in a few weeks..hopefully.
No Frequency Generation: In a linear system, it is not possible for the output to have a maximum frequency greater than that of the input. If the input is band-limited to fn then the output will also be limited to fn Exception: Non-linear systems can introduce new frequency components (such as harmonics), but that is not the case for linear systems. So, here we are reading linear systems, thus the output freq response is limited at max input freq which is Nyquist Freq
Just wanted to add a link to another great video that takes this theory that you tremendously explain to actual C++ implementation. ua-cam.com/video/uNNNj9AZisM/v-deo.html
It’s all coming together now. Thank you so much for these and your thorough examples!
You're welcome mate, thanks for the support. Yea, I think this ties in most of the loose ends from the previous videos.
@@akashmurthy Are you back with more ...
its actually insane how good these videos are, akash!
Thanks a lot Lucas!
The bit at the end showing the coefficient and tap relationship is the highlight of the video :-). For me it tied this whole thing together! Thanks again for the brilliant videos.
Ah! I'm glad that solidified it for you! Seeing the impulse response as the coefficients is a key way for understanding FIR filters I suppose. You're welcome, and thanks for checking it out!
At 11:50 the flow chart shows z-1 for each tap. Does each tap operate on the same sample? I seem to remember it might need to operate on older samples, so z-1, z-2, z-3 and so on.
@@legosteveb a signal which flows through 3 z-1 delay blocks is equivalent to it passing through a singal z-3 delay block. As the signal passes through the circuit, each tap adds a sample of delay, and the same signal is sent to the next tap.
Please make more videos !!!
Thanks so much for the donation! Hope to finish the videos soon!
Your emphasis on intuition is super valuable! I'll be patiently waiting for you to continue the series :)
Thanks a lot! I'm glad it helped. Was a little too busy with work, but I'll be restarting work on this series soon!
I don't have much about this knowledge background but your videos explanation are so intuitive to understand, helps a lot, hope to see your next video soon!
This is exactly what I have been looking for. I got into audio programming recently and I have struggled do find good resources that are understandable at my level of math knowledge. Amazing. Thank you!
Thank you! I'm glad you enjoyed the series, more videos coming soon, as soon as I can get my act together!
The pace is perfect- You are doing great work.Thanks for sharing it.
That's great! Thanks very much for the feedback!
super underrated , keep going , wish you best !
Thanks mate!
Guy, you are amazing ! Great way of explaning the topic step-by-step and great visualization ! Thank you :)
Thanks for checking it out!
I'm learning so much! Wish I had a time machine and you was one of my teachers in the early 80's. Thank you!!!
Amazing video. Even the ending hit hard, because you could see how you would build a parametric eq.
Glad this was helpful! It's kind of difficult to build a parametric EQ with this approach, but you can build a rudimentary one for sure.
Hi! Thanks for this beautiful set of video!
@Akash Murthy, regarding your comment @08:50, a partial answer I can give you is from what I remember of my courses of electronics, especially automation, at school. Most of the time, in analog filters, the transfer function appears to follow the rule that the phase looks like the derivative of the gain. I said most of the time, but I do not guarantee this always applies, besides, sometimes you may have an offset of the phase. Translated, if you have a flat gain response, whether it is 0dB or another value, you should have a 0deg phase dB. If you have a constant -6dB/oct or -20dB/dec slope, you should have a constant (flat) -90deg phase. With a constant -12dB/oct or -40dB/dec you should have a constant (flat) -180deg phase, etc... At the cutoff frequency of a first order, which should be ~3dB, you should have a phase of -45deg. And so on... I can't say whether it applies or not in digital filters, most of all, FIR filters but it may be interesting to find out. @08:50, it appears that this theory applies.
Sorry. What I talked about earlier was essentially without active/discrete components... if you have resistors and capacitors and inductors, it should still be true. However, when you add an op amp, it becomes the higher the frequency (and closer to the 0dB open loop gain), the less true this statement is. I guess that in digital signal it should be the same somehow... the closer to Shannon frequency, the more non linear is the phase even if your gain is flat all the way... I mean, if you have a sampling frequency of 384kHz and handle audio frequencies up to 12kHz, you should be quite fine.
Do you know of free Softwares that I can use to test digital filters?
Do you know how yo calculate coefficients of the transfer function in Z of digital IIR filters like Butterworth, Bessel, Chebychev filters with input parameters such as order of filter, cutoff freq, sampling freq, etc? Thanks.
kudos for making such fine explainer series covering the fundamentals, it helps build intuition and serve as a great introduction to the topic ahead, hope you continue with this work
That's great! Glad you think so.
Really satisfying content here. Glad I found this beautiful little corner of youtube
Thank you!
Thanks @Akash JI for such an amazing explanation with Animation. Every line of explaining you have given in this video is fully useful.
Thank you so so much for colour coding all these different video series
Ah, you're the first to bring up colour coding, thank you for fixating on that! When UA-cam throws random video recommendations down your way, it just makes sense to order similar content by colour.
So it shifts phase to cancel out the signal, and that how its filtered. I would call this Phase Filtering. The delay is relative to the sample rate, filtering the signal. Thank you so much, best tuts ever.
Thank you so much for this incredible tutorial on digital filters! I look forward to future installments. 🙏
Thanks very much! :)
Thanks. Why education system doesn't have professors like you? Thanks a lot brother to you and your team for such wonderful content. Waiting for IIR filter.
Thank you! It's just me..and I'm no professor! Just an enthusiast.
What an awesome video. I absolutely love the conclusion. One of my biggest "aha!" effects ive had in the topic of filters. Also very nice graphics and demonstrations. The FIR Filter plugin part of the viodeo could´ve used a small overview in which all the values are presented at once at the end. For example the combination +0.5 and +0.5, -0.5 and +0.5, -0.5 and -0.5 and so on in one overview. I understood every word, even though im a german. Very good work and a big thank you.
That's awesome to hear, thanks very much for the feedback. Yes, I think I could've spent a little more time showing the plugin , with different values.
@@akashmurthy Update: Got my grade from the exams yesterday and im very happy with the outcome. Thanks to your nice Videos. Good Job :)
@@nonmarking1 well done! :)
Thank you so much for this video. It has helped me tremendously. Greetings from Switzerland :)
@@tinasalvisberg6816 you're very welcome. Cheers from Ireland!
legendary stuff man...thanks a ton for sharing this masterpiece
Good job as always. Thanks 👍
Thanks as always mate!
These videos are great, ive really enjoyed watching them. The production is amazing. How do you do all your animation, camera movement etc? Its really slick.
Thank you! I use Adobe After Effects for he animations and production. Sorry for the late reply.
Admittedly I haven't finished your series of videos and may need to re-watch a few already, but am I correct in thinking feed foward filters are non-resonant? I'm a little curious how resonance works with convolution, but maybe that's explained in another video.
Excellent videos, BTW. Your efforts here are greatly appreciated.
Thanks a lot! I didn't knew we can code DSP in Python, my poor prof. only uses matlab
Thanks for showing an application.
These are fantastic, thank you!
Thanks for checking it out!
Congrats! very good video, animations, presentation, programing and examples. Can I ask you what app or program did use for presentation? Thanks for the video!
Thanks for checking it out! I used After Effects for the animations.
Thank you sooooooo much. Just so clearly explained🙏
You're welcome! Thanks for checking it out
Could you please add videos about Audio codecs like SBC, AAC etc.. sir?
You're the best when it comes to Audio basics :)
Thanks very much! I don't know if I want to do audio codec specific topics, it doesn't interest me very much unfortunately. I'm more interested in general DSP topics.
LEGEND..
Thanks! This videos are made so well!!
When will the next video be uploaded?
Thanks very much! I'm kinda held up with many things at the moment, but hopefully soon! I have the content already, just need to be animated..
@@akashmurthy I am really looking forward to the rest videos of this topic. This is the first time I can understand a little bit concept on DSP. Your videos are super helpful for me. Thank you very much!
@akashmurthy Do we have feedback filter vidoes as well in your channel ?
Not yet, I've been a bit busy over the last year. Haven't got around to it. But I'm making it right now, so it should be out in a few weeks..hopefully.
Sir incredible.
When the feed back videos will come?
Thank you. Soon hopefully..
Sir in quantization noise, is the noise frequency is also within nyquist frequency?
@@ebadurrahmankhan9033 yes, it has to be. There is no way to represent any frequency higher than the Nyquist frequency within the digital domain.
Beautiful!
Do you plan on doing videos on feedback filters as well?
Yea, I've got the script written, just need to find time to animate it.
wow i am using ableton because you can delay 1 sample and i am testing this is so cool.
Happy testing!
No Frequency Generation: In a linear system, it is not possible for the output to have a maximum frequency greater than that of the input. If the input is band-limited to fn then the output will also be limited to fn
Exception: Non-linear systems can introduce new frequency components (such as harmonics), but that is not the case for linear systems.
So, here we are reading linear systems, thus the output freq response is limited at max input freq which is Nyquist Freq
I thought it was just me at first but the repository link leads to a 404 page
Oh sorry, had forgotten to mark it back as public. It should work now.
Just wanted to add a link to another great video that takes this theory that you tremendously explain to actual C++ implementation. ua-cam.com/video/uNNNj9AZisM/v-deo.html
Thanks for the recommendation!
@akashmurthy Do we have feedback filter vidoes as well in your channel ?