Your videos are really helpful for people like me with no basis in audio engineering or anything. Really clear and easy to understand. Don't know why they aren't more popular!
Hey Matt, don't know if you'll ever read this, but every now and then I look at your channel, thinking I'll play a video and see how much I've forgotten, but usually I wind up learning all new concepts. I'm glad you left this channel up. It's good to hear your voice again too! (Don from MN, Bass in BW. Guess who's teaching studio recording to middle schoolers now!)
In 2020, there are many higher-end USB microphones that use 24-bit signals. Your visualization for why the noise floor is important makes analyzing the pros and cons of the additional cost for the 24-bit microphones very easy.
(1/2) I think it's very interesting how many concepts in audio processing often have their counterparts in image processing, this personally helped me to understand things better. For example dithering: When you have a million-colours picture (24bit per pixel) and want to save it as, say 256-colours (8bpp) then using dither makes a huge difference (go to the Wikipedia article on "Dithering" for examples). The other question (recording LP to 24bit) can be explained similarly:
Yes. Different equipment will have different numbers, so those will not always be exactly the case, but in most situations, that's true. Lossless rips are still a good idea, since different people have different hearing and some might hear data compression that others don't, but 24 bits is usually overkill for vinyl. It won't hurt anything, but in that case the information 24 bits hold that 16 bits don't will just be noise.
Yes, you're right - it's considered best practice to leave dither on. That way, there's no chance of any quantization distortion on the fade-outs and other quiet parts - most people agree that quantization distortion sounds a heckuva lot uglier than dither noise. As far as my own work, the only time I ever turn dither off is when using certain test signals where I know I want total digital black (all 0 samples) during silent parts.
The dynamic range over a frequency range of interest also depends on the sampling rate, which distributes the dither noise over a wider band. But quadrupling the sampling rate is more expensive than adding one bit.
Your videos are really helpful for people like me with no basis in audio engineering or anything. Really clear and easy to understand. Don't know why they aren't more popular!
Hey Matt, don't know if you'll ever read this, but every now and then I look at your channel, thinking I'll play a video and see how much I've forgotten, but usually I wind up learning all new concepts. I'm glad you left this channel up. It's good to hear your voice again too! (Don from MN, Bass in BW. Guess who's teaching studio recording to middle schoolers now!)
In 2020, there are many higher-end USB microphones that use 24-bit signals. Your visualization for why the noise floor is important makes analyzing the pros and cons of the additional cost for the 24-bit microphones very easy.
Thank you for uploading these videos! I can't believe that I can learn something like that on the internet for free!
Thanks for the comment!
(1/2)
I think it's very interesting how many concepts in audio processing often have their counterparts in image processing, this personally helped me to understand things better. For example dithering: When you have a million-colours picture (24bit per pixel) and want to save it as, say 256-colours (8bpp) then using dither makes a huge difference (go to the Wikipedia article on "Dithering" for examples).
The other question (recording LP to 24bit) can be explained similarly:
Yes. Different equipment will have different numbers, so those will not always be exactly the case, but in most situations, that's true.
Lossless rips are still a good idea, since different people have different hearing and some might hear data compression that others don't, but 24 bits is usually overkill for vinyl. It won't hurt anything, but in that case the information 24 bits hold that 16 bits don't will just be noise.
Yes, you're right - it's considered best practice to leave dither on.
That way, there's no chance of any quantization distortion on the fade-outs and other quiet parts - most people agree that quantization distortion sounds a heckuva lot uglier than dither noise.
As far as my own work, the only time I ever turn dither off is when using certain test signals where I know I want total digital black (all 0 samples) during silent parts.
Great videos! Could you maybe cover some more of the synthesis part? Sound design and such? Really like your videos!
Thanks! I appreciate it.
The dynamic range over a frequency range of interest also depends on the sampling rate, which distributes the dither noise over a wider band. But quadrupling the sampling rate is more expensive than adding one bit.
True, that's how DSD works to provide a usable dynamic range with just one bit.
Great! 😂