This is great. It’s the most in depth audio series I’ve found. You are great for putting this up for free. You could easily sell this. Thank you so much!
This course is excellent. Your channel rocks. I donno why this is underrated. I recommended this channel to my friends who are interested in these topics. Everything is very technical and precise, yet perfectly thought.
Dude you saved my life! I was going crazy seeing all of those mind numbing videos by musicians explaining that bit depth has no effect on sound whatsoever and that vertical resolution isn't a thing, etc. I really needed to know WHAT happens and understand what I'm doing, and this series of videos provided that.
Great Videos! Glad to be under the first 500 subscribers to this channel. It's going to be well known channel among the students for sure! Well done Sir!
Hey Aditya, thanks for checking it out. Here are a few resources that I refer to: digitalsoundandmusic.com/ - Pretty amazing source, and it's completely free. jackschaedler.github.io/circles-sines-signals/index.html - A good place to check some basic math and see some cool visualization Books: Ken Pohlmann - Principles of Digital Audio (Super gritty and detailed, sometimes a little too complicated) David Howard - Acoustics and Psychoacoustics (Great handbook for audio in the physical domain) Richard Boulanger - The Audio Programming Book (Super detailed book for programming principles for audio, though everything's written in C and a bit janky. The principles are sound however)
No, it's just a plain sum. Maybe the graphics were a bit misleading? If you want to get a triangular PDF between 0 and 100, you add a random value between (0,50) to another independent random value between (0,50). tri = random(50) + random(50) No expensive convolution required.
@@akashmurthy When the two uniform random variables are added like you mentioned you get a new random variable whose pdf is convolution of the individual pdfs. ie you get a triangular signal when you convolve two rectangular signals.
@@shyamm4680 you're technically right. en.m.wikipedia.org/wiki/Convolution_of_probability_distributions#:~:text=The%20convolution%20of%20probability%20distributions,linear%20combinations%20of%20random%20variables. I was thinking of the convolution of 2 time domain signals, which is an integral of the product of the 2 function. In this case it's a sum.
Hello! I would like to be sure of the following: Rectangular pdf gets worse SQNR but better to uncorelated from the signal (no distortion), Triangular pdf gets better SQNR than rectangular pdf (but a bit worse about the uncorelated comparing with rectangular ?) and... Gaussian gets better SQNR than both 2 but worse about uncorelated the signal (more distortion than both even if it's minimal) ? Thanks !
Sir as we tend to use High Bit Depth like 24 Bits of 16 bits, we need a lot of levels, Now todays technology has mostly VDD at 1.8V So in between them how do we accommodate such a huge number of bit dept? Because the difference between two Bit will be so narrow. How the engineers are designing their ADC or DAC for that?
Haha, I'm an audio engineer, not a graphic designer ;) You're right, the colour banding on gradients is really annoying. I've tried adding a bit of noise to the background, but it makes the file size really huge and it makes the video fuzzy to watch. I haven't figured out a good way to add noise for this problem. If I use 16-bit colour instead of 8-bit, the banding goes away. But once I upload it to TH-cam, its compression algorithm brings the banding back. Thoughts?
@@akashmurthy the fault is TH-cam's, I guess. I know for a fact that uploading a video in 4K instead of 1080p forces YT to allocate more bitrate to it, thus reducing extreme artifacts - but will it help with banding? Until the world of video streaming and displaying becomes universally 10-bit, that irritating banding may never go away...
Great video! I didn't get how the formation of correlated distortion during bit depth reduction and non-linearity on the input output characteristics graph are related and couldn't find good material on the Internet for further and more detailed reading on the subject. Do you have any suggestions? Thank you!
Cheers! For a deeper dive, I'd definitely recommend: Principles of Digital Audio by Ken C. Pohlmann Another reference that I used: ftp.sea-acustica.es/fileadmin/publicaciones/Sevilla02_spagen001.pdf
This is great. It’s the most in depth audio series I’ve found. You are great for putting this up for free. You could easily sell this. Thank you so much!
Arguably the best series on digital audio on youtube! Keep it up brother im sure that u will get more recognition for your work after some time
Thank you so much my dude!
This course is excellent. Your channel rocks. I donno why this is underrated. I recommended this channel to my friends who are interested in these topics. Everything is very technical and precise, yet perfectly thought.
Thank you so much for your comment, and for sharing it around :)
Thank you so much for this super awesome series of educational videos about DSP!!! 🙏💜
You're welcome, thanks for checking it out.
Excellent explanation of the probability density function
Hey thanks very much! I'm glad you found it useful! Not a lot of people find PDFs useful.
Dude you saved my life! I was going crazy seeing all of those mind numbing videos by musicians explaining that bit depth has no effect on sound whatsoever and that vertical resolution isn't a thing, etc. I really needed to know WHAT happens and understand what I'm doing, and this series of videos provided that.
Thank you! I'm glad you found the series useful!
Damn the amount of detail and quality of this series is phenomenal, thanl-you so much!
Oh, thanks very much for checking it out!
Great Videos! Glad to be under the first 500 subscribers to this channel. It's going to be well known channel among the students for sure!
Well done Sir!
Thank you for your confidence in the channel :)
Hey! Love your videos. Can you please suggest some resources, books or websites, where I can read up on this more?
Hey Aditya, thanks for checking it out.
Here are a few resources that I refer to:
digitalsoundandmusic.com/ - Pretty amazing source, and it's completely free.
jackschaedler.github.io/circles-sines-signals/index.html - A good place to check some basic math and see some cool visualization
Books:
Ken Pohlmann - Principles of Digital Audio (Super gritty and detailed, sometimes a little too complicated)
David Howard - Acoustics and Psychoacoustics (Great handbook for audio in the physical domain)
Richard Boulanger - The Audio Programming Book (Super detailed book for programming principles for audio, though everything's written in C and a bit janky. The principles are sound however)
@@akashmurthy Thanks a lot! You are a blessing!
All the modules are great and clear. The only thing I didn’t quite understand is if the scales are actually LSB or is some range of LSB normalized.
Keyword 'convolution' instead of 'sum' at 2:30min?
No, it's just a plain sum. Maybe the graphics were a bit misleading? If you want to get a triangular PDF between 0 and 100, you add a random value between (0,50) to another independent random value between (0,50).
tri = random(50) + random(50)
No expensive convolution required.
@@akashmurthy When the two uniform random variables are added like you mentioned you get a new random variable whose pdf is convolution of the individual pdfs. ie you get a triangular signal when you convolve two rectangular signals.
@@shyamm4680 you're technically right.
en.m.wikipedia.org/wiki/Convolution_of_probability_distributions#:~:text=The%20convolution%20of%20probability%20distributions,linear%20combinations%20of%20random%20variables.
I was thinking of the convolution of 2 time domain signals, which is an integral of the product of the 2 function.
In this case it's a sum.
Hello! I would like to be sure of the following:
Rectangular pdf gets worse SQNR but better to uncorelated from the signal (no distortion),
Triangular pdf gets better SQNR than rectangular pdf (but a bit worse about the uncorelated comparing with rectangular ?) and...
Gaussian gets better SQNR than both 2 but worse about uncorelated the signal (more distortion than both even if it's minimal) ?
Thanks !
Sir as we tend to use High Bit Depth like 24 Bits of 16 bits, we need a lot of levels,
Now todays technology has mostly VDD at 1.8V
So in between them how do we accommodate such a huge number of bit dept?
Because the difference between two Bit will be so narrow.
How the engineers are designing their ADC or DAC for that?
I don't know! Just extremely sensitive electronic components I guess! Not my area of expertise..
Thanks for this great tutorial! May I ask what's the software tool you used to generate such great animation?
Thank you! I use After Effects for the animations..
the background gradient image suffers from quantization errors. It needs to be dithered...
Haha, I'm an audio engineer, not a graphic designer ;)
You're right, the colour banding on gradients is really annoying. I've tried adding a bit of noise to the background, but it makes the file size really huge and it makes the video fuzzy to watch. I haven't figured out a good way to add noise for this problem. If I use 16-bit colour instead of 8-bit, the banding goes away. But once I upload it to TH-cam, its compression algorithm brings the banding back. Thoughts?
@@akashmurthy the fault is TH-cam's, I guess. I know for a fact that uploading a video in 4K instead of 1080p forces YT to allocate more bitrate to it, thus reducing extreme artifacts - but will it help with banding? Until the world of video streaming and displaying becomes universally 10-bit, that irritating banding may never go away...
Great video! I didn't get how the formation of correlated distortion during bit depth reduction and non-linearity on the input output characteristics graph are related and couldn't find good material on the Internet for further and more detailed reading on the subject. Do you have any suggestions? Thank you!
Cheers! For a deeper dive, I'd definitely recommend: Principles of Digital Audio
by Ken C. Pohlmann
Another reference that I used: ftp.sea-acustica.es/fileadmin/publicaciones/Sevilla02_spagen001.pdf
@@akashmurthy Thanks for the link! I also had trouble understanding that part of the video. Figure 2 of that pdf really helped.