|
|
Post by geoff738 on Jun 11, 2024 15:20:31 GMT -6
Ok, these probably aren’t inherent to Logic, but anyways. I have left the bit rate at 24 rather than 32 bit floating point. My understanding is it doesn’t really matter, as long as you aren’t clipping the stereo outs. But is that correct? And dithering before sending out to hardware. I know we had a lengthy thread about dither a couple or three months back, but it obviously didn’t sink in for me. Am I correct that the dither goes right after the i/o plug? Does it matter if I engage delay compensation before or after inserting the dither plug? And then how to set the dither plug. I am using good dither. dan? Cheers, Geoff |
|
|
|
Post by kcatthedog on Jun 11, 2024 15:32:06 GMT -6
I thought 32 bit gives you the lowest noise floor and highest headroom.
Unless, you are reducing the bit rate why are you dithering?
|
|
|
|
Post by nudwig on Jun 11, 2024 15:34:25 GMT -6
My rule of thumb is anytime it leaves the DAW to the real world (inserts and final bounce) it needs to be dithered to 24 bit. In your case that'd be before the I/O plugin.
|
|
|
|
Post by kcatthedog on Jun 11, 2024 16:16:48 GMT -6
But why, if you are not truncating bits?
|
|
|
|
Post by bgrotto on Jun 11, 2024 16:24:05 GMT -6
Just gotta be “that guy” and point out there’s a difference between bit rate and bit depth. You guys are discussing the latter.
Higher bit depth doesn’t create headroom on the analog side; that’s a fixed value. It does create more “foot room”if you like, allowing for higher resolution capture of quieter signals. Where you gain headroom is in digital calculations, and 32 bit provides basically infinite headroom there. But you still gotta trim the output down if you’re exceeding 0dbfs before hitting a DA.
I’ll just see myself out now…..
|
|
|
|
Post by geoff738 on Jun 11, 2024 17:08:37 GMT -6
Yeah I guess my confusion lies in the contention that you should always dither when going out of your Daw. But does that hold true if you are not altering the bit, ahem, depth? Thanks for the correction Benny!
Would appreciate someone setting me straight on this.
Cheers,
Geoff
|
|
|
|
Post by kcatthedog on Jun 11, 2024 17:19:49 GMT -6
That’s all I was asking about: have always understood 32 is better than 24 for lower noise floor and that unless reducing the length of the bit word, you don’t need to dither or you may be able to measure but not hear the difference ?
|
|
|
|
Post by wiz on Jun 11, 2024 17:24:45 GMT -6
Yeah I guess my confusion lies in the contention that you should always dither when going out of your Daw. But does that hold true if you are not altering the bit, ahem, depth? Thanks for the correction Benny! Would appreciate someone setting me straight on this. Cheers, Geoff Back when I used LOGIC and and outboard.....pehaps logic 10......After a conversation with Bob O.... I placed air windows dither in the plug in slot before the IO plug in in logic..... I agree with both previous posts, when going out of the DAW analog.....dither.... when reducing bit depth ...dither.... increased bit depth does not provide analog headroom... that is limited physically by the audio interface itself...what it does do is provide foot room... Look, it's not something that is going to make or break a production.. but something to be aware of and try for yourself and see if it matters to you cheers Wiz |
|
|
|
Post by wiz on Jun 11, 2024 17:29:25 GMT -6
That’s all I was asking about: have always understood 32 is better than 24 for lower noise floor and that unless reducing the length of the bit word, you don’t need to dither or you may be able to measure but not hear the difference ? Noise floor is a function of the original conversion of audio to digital.....when tracking the original source.....or once in the DAW.. the processing you are doing. cheers Wiz |
|
|
|
Post by the other mark williams on Jun 11, 2024 21:33:27 GMT -6
Ok, these probably aren’t inherent to Logic, but anyways. I have left the bit rate at 24 rather than 32 bit floating point. My understanding is it doesn’t really matter, as long as you aren’t clipping the stereo outs. But is that correct? And dithering before sending out to hardware. I know we had a lengthy thread about dither a couple or three months back, but it obviously didn’t sink in for me. Am I correct that the dither goes right after the i/o plug? Does it matter if I engage delay compensation before or after inserting the dither plug? And then how to set the dither plug. I am using good dither. Cheers, Geoff There are far more intelligent people about this matter here on RGO than me, but hey, I've been up for 36hrs now and so it looks like I'm taking a swing at it. Smart people, if I've misstated something below, please correct me. I'm here to learn. There are two distinct concepts here that seem to be getting mixed up: 1. the audio engine of your DAW (in this case, Logic's mixer), and 2. the audio files themselves which are recorded into Logic via your converters/interface (or imported into your session, but recorded somewhere else outside of Logic). Re: the audio engine, Logic has had a 32-bit floating point mixer for many years now - that is, the processing happens as 32-bit floating point. But in regards to audio files, Logic has only been able to directly work with 32-bit float audio files for like maybe a year or so. (Whereas Reaper, for one, has been able to do so for quite a few years.) If you have 24-bit converters--and you almost certainly do--you're not really recording an actual 32-bit float WAV file--at least not with the actual benefits of a 32-bit float file. So whenever you leave Logic's mixer (which operates at 32-bit float), you're truncating it down to 24-bit because that's what your converter can actually output. So conventional best practice is to dither before sending the audio out. (There is a lively debate about whether we can actually hear the difference between a 32-bit file dithered to 24-bit vs. truncated to 24-bit. I am not taking sides in that debate for the purpose of this post.) Back to the audio files themselves: Why even have Logic set to use 32-bit float files? Well, in my case, I frequently use recorders in the field that actually do record 32-bit float audio files (Sound Designs MixPre series and Zoom F6 in my case). Prior to Logic supporting 32-bit float files, it would just truncate those files to 24-bit at import, destroying the benefit of the recording method I had chosen. So I had to edit with something different. (If I'm recording in the studio with my Metric Halo ULN-8, it doesn't matter whether Logic is set to record 24-bit files or 32-bit float files, as the MH is a 24-bit A/D.) But why, if you are not truncating bits? But when you leave the DAW to go to your outboard, you are truncating bits, because the audio engine is operating at 32-bit float, but your converters are only 24-bit. Yeah I guess my confusion lies in the contention that you should always dither when going out of your Daw. But does that hold true if you are not altering the bit, ahem, depth? Thanks for the correction Benny! Would appreciate someone setting me straight on this. Cheers, Geoff See above. Hope it's a little clearer... |
|
|
|
Post by Dan on Jun 11, 2024 22:44:47 GMT -6
Ok, these probably aren’t inherent to Logic, but anyways. I have left the bit rate at 24 rather than 32 bit floating point. My understanding is it doesn’t really matter, as long as you aren’t clipping the stereo outs. But is that correct? And dithering before sending out to hardware. I know we had a lengthy thread about dither a couple or three months back, but it obviously didn’t sink in for me. Am I correct that the dither goes right after the i/o plug? Does it matter if I engage delay compensation before or after inserting the dither plug? And then how to set the dither plug. I am using good dither. Cheers, Geoff There are far more intelligent people about this matter here on RGO than me, but hey, I've been up for 36hrs now and so it looks like I'm taking a swing at it. Smart people, if I've misstated something below, please correct me. I'm here to learn. There are two distinct concepts here that seem to be getting mixed up: 1. the audio engine of your DAW (in this case, Logic's mixer), and 2. the audio files themselves which are recorded into Logic via your converters/interface (or imported into your session, but recorded somewhere else outside of Logic). Re: the audio engine, Logic has had a 32-bit floating point mixer for many years now - that is, the processing happens as 32-bit floating point. But in regards to audio files, Logic has only been able to directly work with 32-bit float audio files for like maybe a year or so. (Whereas Reaper, for one, has been able to do so for quite a few years.) If you have 24-bit converters--and you almost certainly do--you're not really recording an actual 32-bit float WAV file--at least not with the actual benefits of a 32-bit float file. So whenever you leave Logic's mixer (which operates at 32-bit float), you're truncating it down to 24-bit because that's what your converter can actually output. So conventional best practice is to dither before sending the audio out. (There is a lively debate about whether we can actually hear the difference between a 32-bit file dithered to 24-bit vs. truncated to 24-bit. I am not taking sides in that debate for the purpose of this post.) Back to the audio files themselves: Why even have Logic set to use 32-bit float files? Well, in my case, I frequently use recorders in the field that actually do record 32-bit float audio files (Sound Designs MixPre series and Zoom F6 in my case). Prior to Logic supporting 32-bit float files, it would just truncate those files to 24-bit at import, destroying the benefit of the recording method I had chosen. So I had to edit with something different. (If I'm recording in the studio with my Metric Halo ULN-8, it doesn't matter whether Logic is set to record 24-bit files or 32-bit float files, as the MH is a 24-bit A/D.) But why, if you are not truncating bits? But when you leave the DAW to go to your outboard, you are truncating bits, because the audio engine is operating at 32-bit float, but your converters are only 24-bit. Yeah I guess my confusion lies in the contention that you should always dither when going out of your Daw. But does that hold true if you are not altering the bit, ahem, depth? Thanks for the correction Benny! Would appreciate someone setting me straight on this. Cheers, Geoff See above. Hope it's a little clearer... all 24-bit fixed numbers correspond to a 32-bit float number but the reverse isn't true so you must use dither to make the conversion linear with a noise floor. the modern converters don't even hit 24-bits and have dither noise higher than that, usually 18 to 21 bits but some are worse. 24-bit is just another byte over 16.
|
|
|
|
Post by kcatthedog on Jun 12, 2024 10:43:14 GMT -6
|
|
|
|
Post by geoff738 on Jun 13, 2024 11:22:31 GMT -6
Ok the mastering the mix says you don’t need to dither from 32 bit floating point since its effectively 24 bit leaving the Daw.
I suspect I will have to experiment with and without dither and draw my own conclusions.
Cheers,
Geoff
|
|
|
|
Post by Dan on Jun 14, 2024 0:34:13 GMT -6
serious misinformation in that article from programmers or pr guy of stupid redundant plugins Random points I’ll make before I got sick of the article Dither does not reduce quantization distortion; it eliminates it. Digital is not capturing a snapshot. It is not discrete snapshots! It is the signal band limited with a noise floor when reconstructed with the anti-alias filter. Of course in the real world it must pass through filters and analog amplifiers that affect the impulse response and add minute (hopefully) distortion. Bit depth does not convey resolution! It only conveys the noise floor. There is information beneath, within the dither noise floor in digital audio. Higher sample rates do not capture more information. You cannot hear that high! The nyquist limit determines the highest frequencies it can capture. Continuous time signals varying in speed are not pixel counts. The dynamic range of real world audio is far less than 24-bit or even 16-bit. Quiet mic amplified 50 db by a quiet pre already has a lot of noise and the ambient noise within it will be higher than that. You cannot record with 32-bit fixed resolution. Modern converters Are 20-bit at best if you spend a lot of money. 24-bit is technically impossible given the Johnson noise of a single resistor at 1 kHz. Dither cannot prevent rounding errors which occur in floating point audio. Low bit depth, high sample rate is how modern converters work with delta sigma modulation. It works better than resistor ladders in the real world. Dither is always necessary from floating point to fixed. 32-bit float numbers do not map to 24-bit fixed ones. Your converters are all ultimately fixed point so dithering should be before the converters as your last monitoring insert. Flat dither is better than noise shaped for real world audio bit depths of 24-bit and 16-bit Sending your mastering guy a 32-bit float file will not increase the detail versus a 24-bit or even a 16-bit. 16-bit dithered does nothing but have a higher noise floor. |
|
