In short, 24 bit and 32 bit are identical audio, they null when inverted. The advantage to 32 bit floating point is it stores the data above digital 0dB to help save a clipping file by later being able to reduce the gain. Please check out my previous video on bit depth for a deeper explanation. Thanks for the question! th-cam.com/video/m4-V_WnV6LE/w-d-xo.html
Oversampling isn't automatic bettermaker you are smearing the transients when using tons of oversampled plugins with those anti aliasing filters and also it inevitably changes the way detector circuit of the compressor works. Also the EQ will not cramp at near niquist frequency when boosting the "air" frequencies of the instrument because there will be sufficient headroom above most eqs bands capabilities. In fact the best choice if ultimate quality is what is desirable is to use 192KHz or more (ofcourse that's more CPU intensive) and then use the filters only when needed above hearable range yourself using something like free Airwindows hypersonic filter (steep filter above 25KHz to minimise phase effect below 20KHz) or something else where you have total control over the steepness and phase response of that filter. This way all plugins (except guitar pedals and heavy distortion plugins) will run at 0 samples latency and sound closer to the hardware that they're emulating. Also I like to use that SSL bus comp 2 without Oversampling because my go to settings on mix bus are 1.5:1 ratio, 30ms attack and auto or medium quick release where the generated aliasing is 90dB below main signal in worst case so fully masked, and that 2x and 4x doesn't quite sound right to me at those settings it just looses bit of punch and natural drive and aliveness that SSL compressor is so famous for 🙂 Cheers!
Fair enough! It's a lot of moving parts and the more parts the more complicated which is why I was trying to find the Goldilocks Zone. The setup that makes the most sense given all the pros and cons. I find when you weigh it all out, 48khz makes more sense, but of course it's not the highest possible fidelity you can achieve. But saving the CPU when running a session makes more sense to me than going for a resolution that makes fractional improvements. Thanks for taking the time to answer so thoughtfully and thoroughly. Much appreciated
Dude, you should make more videos on this subject. You're great at explaining this stuff. Thanks.
Thanks! I hope to keep creating monthly production tip videos like this.
Make more videos man about making music , you're a talent ! Cheers !
Thank you! I really appreciate that. More videos are in the works so I hope you subscribe!
Can you go a bit further and explain why the 32-bit floating point is ideal? Thanks.
In short, 24 bit and 32 bit are identical audio, they null when inverted. The advantage to 32 bit floating point is it stores the data above digital 0dB to help save a clipping file by later being able to reduce the gain. Please check out my previous video on bit depth for a deeper explanation. Thanks for the question! th-cam.com/video/m4-V_WnV6LE/w-d-xo.html
Oversampling isn't automatic bettermaker you are smearing the transients when using tons of oversampled plugins with those anti aliasing filters and also it inevitably changes the way detector circuit of the compressor works. Also the EQ will not cramp at near niquist frequency when boosting the "air" frequencies of the instrument because there will be sufficient headroom above most eqs bands capabilities.
In fact the best choice if ultimate quality is what is desirable is to use 192KHz or more (ofcourse that's more CPU intensive) and then use the filters only when needed above hearable range yourself using something like free Airwindows hypersonic filter (steep filter above 25KHz to minimise phase effect below 20KHz) or something else where you have total control over the steepness and phase response of that filter.
This way all plugins (except guitar pedals and heavy distortion plugins) will run at 0 samples latency and sound closer to the hardware that they're emulating.
Also I like to use that SSL bus comp 2 without Oversampling because my go to settings on mix bus are 1.5:1 ratio, 30ms attack and auto or medium quick release where the generated aliasing is 90dB below main signal in worst case so fully masked, and that 2x and 4x doesn't quite sound right to me at those settings it just looses bit of punch and natural drive and aliveness that SSL compressor is so famous for 🙂
Cheers!
Fair enough! It's a lot of moving parts and the more parts the more complicated which is why I was trying to find the Goldilocks Zone. The setup that makes the most sense given all the pros and cons. I find when you weigh it all out, 48khz makes more sense, but of course it's not the highest possible fidelity you can achieve. But saving the CPU when running a session makes more sense to me than going for a resolution that makes fractional improvements. Thanks for taking the time to answer so thoughtfully and thoroughly. Much appreciated