|
Post by gouge on Dec 11, 2016 6:49:04 GMT -6
never thought i'd say I like the sound of a dither plugin. up till now they've mostly been there and not really heard.
airwindows pauldither changed that. I like the sound of it a lot. to describe the sound.
it is smooth, open, airy and clear. it seems to have taken me a significant step towards removing the "sound" that I didn't like about digital.
|
|
|
Post by svart on Dec 11, 2016 9:04:56 GMT -6
iZotope 64 bit ultra SRC dither has won multiple times over everything else I've tried over the years. It's the closest thing to sounding like 24/88.2k after being converted to 44.1k that I've heard.
|
|
|
Post by Johnkenn on Dec 11, 2016 11:42:06 GMT -6
I still don't understand all this shit.
|
|
|
Post by svart on Dec 11, 2016 12:48:40 GMT -6
I still don't understand all this shit. You record at a higher bit depth and sample rate for maximum fidelity, like 24/96. However, you need to deliver a mix at a lower bit depth and sample rate, i.e., 16/44.1. You could redo your mixes after switching your converters from 24/96 to 16/44.1, or you could run them though a converting program. So you choose to convert them, but in the digital domain, for very nerdy technical reasons, you cant simply chop off the 8 bits you need to get rid of to go from 24 down to 16. You'd end up having horrible noise if you did it by simply cutting out those bits. Likewise, if you did it other ways like rounding bits in one direction or another, you'd get various forms of harmonic trash, or other distortions that arise from the actions. The best way to do it, is to do random rounding of the bits to not only reduce the number of bits, but to randomize the distortions that might arise from the process. The human ear and brain are very good at hearing repeating tones, but very bad at picking out random tones. Dither simply randomizes the distortions and artifacts from rounding down bits so that the brain is fooled into ignoring them.
|
|
|
Post by Johnkenn on Dec 11, 2016 14:14:20 GMT -6
Well, I understand what dither is and does. I was talking more about what Bob was talking about - dithering every channel and all that jazz. I usually record at 24/48 and stay in that when mixing. Then I go out to outboard and summing (I have 24 bit dither engaged) and record back in. Then I export the final 24/48 wav. If I need an MP3, I encode in iTunes. If someone needs to put it on a cd, I let them deal with it or mastering deal with it. Haven't had to reduce bit rate in a couple of years.
|
|
|
dither
Dec 11, 2016 14:26:37 GMT -6
Post by svart on Dec 11, 2016 14:26:37 GMT -6
Well, I understand what dither is and does. I was talking more about what Bob was talking about - dithering every channel and all that jazz. I usually record at 24/48 and stay in that when mixing. Then I go out to outboard and summing (I have 24 bit dither engaged) and record back in. Then I export the final 24/48 wav. If I need an MP3, I encode in iTunes. If someone needs to put it on a cd, I let them deal with it or mastering deal with it. Haven't had to reduce bit rate in a couple of years. I guess if you import a project at a higher bit depth than your hardware can handle (but what hardware made in the last decade can't do 24 bit?), you'd dither every channel down.. But math precision is much better(but slower) at higher bit depths and sample rates, so you'd want to work at as high of a bit depth as possible and leave dither to the last process.
|
|
|
Post by EmRR on Dec 11, 2016 14:33:24 GMT -6
I'm almost always tracking 88.2/24 and mixing analog back to 88.2/32 float, so individual DAC outputs have 24 bit flat dither (post-processing/fader), as well as monitoring output. Clients always have to receive down-converted reference copies, letting DP do the 88.2/32 -> 44.1/16 part and iTunez doing the mp3 (320kbps VBR highest quality) part. I have messed with the LAME framework a little, and so far haven't made an mp3 myself that sounded as 'correct' as what itunez does. 88.2/32 goes to mastering guy, they pick final dither type. If I attend, I can usually hear it, but don't live with any one long enough to ever develop a definite preference.
|
|
|
dither
Dec 11, 2016 15:22:07 GMT -6
Post by Johnkenn on Dec 11, 2016 15:22:07 GMT -6
So - you guys think the 32 bit float (wth does all that mean) sounds better?
|
|
|
Post by Bob Olhsson on Dec 11, 2016 15:36:49 GMT -6
You'd never use dither for recording unless you are recording to 16 bit files.
When you need dither is when audio is being processed because that expands the bit depth to as close to infinity as the hardware and software allow and the resulting number stream needs to be dithered down to 24 bits so it can be streamed to a D to A or written to a file. This prevents distortion that masks low level information and builds up to crunchy sound. The one exception is printing a floating point file which minimizes but does not completely eliminate the distortion.
Airwindows is real good but used to only be audioUnits so I'd never tried it. The new shareware versions are also mac/windoze VST and pauldither sounds really good to me too.
|
|
|
dither
Dec 11, 2016 16:55:52 GMT -6
via mobile
Post by gouge on Dec 11, 2016 16:55:52 GMT -6
Well, I understand what dither is and does. I was talking more about what Bob was talking about - dithering every channel and all that jazz. I usually record at 24/48 and stay in that when mixing. Then I go out to outboard and summing (I have 24 bit dither engaged) and record back in. Then I export the final 24/48 wav. If I need an MP3, I encode in iTunes. If someone needs to put it on a cd, I let them deal with it or mastering deal with it. Haven't had to reduce bit rate in a couple of years. I started using either as per bob's comments a few years back. I mix down via outboard and use itb for combining mics and some automation. I record at 24bit and play back 24bit. I use dither across every track as the last plug before hitting the converters on each track.
|
|
|
dither
Dec 11, 2016 17:22:34 GMT -6
Post by wiz on Dec 11, 2016 17:22:34 GMT -6
I'm almost always tracking 88.2/24 and mixing analog back to 88.2/32 float, so individual DAC outputs have 24 bit flat dither (post-processing/fader), as well as monitoring output. Clients always have to receive down-converted reference copies, letting DP do the 88.2/32 -> 44.1/16 part and iTunez doing the mp3 (320kbps VBR highest quality) part. I have messed with the LAME framework a little, and so far haven't made an mp3 myself that sounded as 'correct' as what itunez does. 88.2/32 goes to mastering guy, they pick final dither type. If I attend, I can usually hear it, but don't live with any one long enough to ever develop a definite preference. How do you insert the dither plug in post fader? Which Daw are you using....? This is always the part of the equation that never gets a clear answer... cheers Wiz
|
|
|
dither
Dec 11, 2016 17:24:55 GMT -6
wiz likes this
Post by EmRR on Dec 11, 2016 17:24:55 GMT -6
I'm using DP. You can set inserts point to be pre or post fader.
|
|
|
dither
Dec 11, 2016 17:35:52 GMT -6
Post by wiz on Dec 11, 2016 17:35:52 GMT -6
I'm using DP. You can set inserts point to be pre or post fader. thanks for that... I am gonna have to do a test on this... cheers Wiz
|
|
|
dither
Dec 11, 2016 21:06:51 GMT -6
wiz likes this
Post by gouge on Dec 11, 2016 21:06:51 GMT -6
wiz, on reaper i use a work around by using folder channels. i apply the dither plug to the folder channel. you can also use gain plugs on your channels and not use the daw faders. i notice chris at airwindows has a new dither plug out. spatialize dither. will need to try this one.
|
|
|
dither
Dec 11, 2016 22:44:35 GMT -6
wiz likes this
Post by popmann on Dec 11, 2016 22:44:35 GMT -6
This is always the part of the equation that never gets a clear answer... cheers Wiz In Cubase you use the last two inserts, which are different colored as they are post fader....
Just top chime in with knowledge unhelpful to your specific question.
But, in general, I don't know why you're using an analog mixer but not summing on it, since that's what 100% broken in floating point software mixers, IMO....I only mention it because my dither plug ins are full time in my template on the outputs of Cubase--otherwise the things that feed the RME mixer. It doesn't matter how I want to do sub bussing--they all remain 32bit float until they are output to a 24bit entity--the RME's hardware mixer or it's DACs.
That's really the easy way to remember "the rule" of dither--your mixer is ALWAYS floating point (32 or 64 depending on app)...any time you want to output to a 24 bit entity--DAC, hardware mixer, or file to be consumed....you need to dither.
You capture 24bits. All you need to write to the disk....it's ALWAYS a floating point (32/64) mixer and plug ins....so, when you do anything and need it to be 24 (or 16bit) again, insert the proper dither.
I'd ask Svart if he's ALSO doing a SRC in Izotope. IME, dither algorithms make very little difference, sample rate makes a TON---meaning if you have a 24/48 (is actuall worst case--not 96) and need 16/44, the dither is relatively trivial compare to what algo you use for SRC. So, if one is doing a 24/96 recording and wanting 16/44 mp3, dither should BE there, but which algo matter more on the SRC than the bit reduction. I understand their SRC is killer, Izotope that is....I've not used it. I feel like I have RX4, I COULD....but, it never occurs to me to open the forensic restoration software to reduce the file size....
|
|
|
Post by popmann on Dec 11, 2016 22:48:21 GMT -6
And no, JK....unless you're recording THROUGH 32bit plug ins, there's no advantage in TRACKING at 32bit float. That said, once you get into editing, I will point out that Cubase can flip back and forth without touching the existing files....so, in a theoretically perect universe, you track with the project in 24bit...flip it to 32bit for whatever editing/DSP you need to do...and if you're ITB, capture the mix at 32....but, you're not--so you need to dither to 24 and capture thyat at 24, so BACK to 24 when you do final mix.
The crux of that is that, in Cubase, the project can flip back and forth without converting existing files. Thus--track at 24....edit and mix at 32...and the only "50% bigger files" are the ones you edited and the mix buss stereo.
(external)Summing users should just leave it on 24. It's easier.
|
|
|
dither
Dec 11, 2016 23:18:39 GMT -6
Post by mrholmes on Dec 11, 2016 23:18:39 GMT -6
In Cubase you use the last two inserts, which are different colored as they are post fader....
Just top chime in with knowledge unhelpful to your specific question.
But, in general, I don't know why you're using an analog mixer but not summing on it, since that's what 100% broken in floating point software mixers, IMO....I only mention it because my dither plug ins are full time in my template on the outputs of Cubase--otherwise the things that feed the RME mixer. It doesn't matter how I want to do sub bussing--they all remain 32bit float until they are output to a 24bit entity--the RME's hardware mixer or it's DACs.
That's really the easy way to remember "the rule" of dither--your mixer is ALWAYS floating point (32 or 64 depending on app)...any time you want to output to a 24 bit entity--DAC, hardware mixer, or file to be consumed....you need to dither.
You capture 24bits. All you need to write to the disk....it's ALWAYS a floating point (32/64) mixer and plug ins....so, when you do anything and need it to be 24 (or 16bit) again, insert the proper dither.
I'd ask Svart if he's ALSO doing a SRC in Izotope. IME, dither algorithms make very little difference, sample rate makes a TON---meaning if you have a 24/48 (is actuall worst case--not 96) and need 16/44, the dither is relatively trivial compare to what algo you use for SRC. So, if one is doing a 24/96 recording and wanting 16/44 mp3, dither should BE there, but which algo matter more on the SRC than the bit reduction. I understand their SRC is killer, Izotope that is....I've not used it. I feel like I have RX4, I COULD....but, it never occurs to me to open the forensic restoration software to reduce the file size....
If I use the desk I always use software monitoring and I dither the input of the rme as mentoined by BK on GS several times....to my ear it works the same as dithering the single outs???
|
|
|
dither
Dec 12, 2016 8:45:37 GMT -6
Post by svart on Dec 12, 2016 8:45:37 GMT -6
In Cubase you use the last two inserts, which are different colored as they are post fader....
Just top chime in with knowledge unhelpful to your specific question.
But, in general, I don't know why you're using an analog mixer but not summing on it, since that's what 100% broken in floating point software mixers, IMO....I only mention it because my dither plug ins are full time in my template on the outputs of Cubase--otherwise the things that feed the RME mixer. It doesn't matter how I want to do sub bussing--they all remain 32bit float until they are output to a 24bit entity--the RME's hardware mixer or it's DACs.
That's really the easy way to remember "the rule" of dither--your mixer is ALWAYS floating point (32 or 64 depending on app)...any time you want to output to a 24 bit entity--DAC, hardware mixer, or file to be consumed....you need to dither.
You capture 24bits. All you need to write to the disk....it's ALWAYS a floating point (32/64) mixer and plug ins....so, when you do anything and need it to be 24 (or 16bit) again, insert the proper dither.
I'd ask Svart if he's ALSO doing a SRC in Izotope. IME, dither algorithms make very little difference, sample rate makes a TON---meaning if you have a 24/48 (is actuall worst case--not 96) and need 16/44, the dither is relatively trivial compare to what algo you use for SRC. So, if one is doing a 24/96 recording and wanting 16/44 mp3, dither should BE there, but which algo matter more on the SRC than the bit reduction. I understand their SRC is killer, Izotope that is....I've not used it. I feel like I have RX4, I COULD....but, it never occurs to me to open the forensic restoration software to reduce the file size....
I use Izotope's other plugin for SRC. They came as a set. I like it too. It's a little slow, but doing fast A/B tests, it works better than most others I've tried. What I mainly listen for is comparison between a natural 44.1, an 88.2/96 file and one that's been downconverted. I find that the Izotope SRC downconverted file sounds more like the 88/96 file than the native 44.1 file does. That's what sold me on it.
|
|
|
dither
Dec 12, 2016 13:59:56 GMT -6
Post by Guitar on Dec 12, 2016 13:59:56 GMT -6
Great thread. I've just added some default 24 bit dither to my main project template. I'll have to remember to dither my hardware insert DAC/ADC loops next time I use them, or the summing mixer.
|
|
|
dither
Dec 12, 2016 14:23:01 GMT -6
via mobile
wiz likes this
Post by ChaseUTB on Dec 12, 2016 14:23:01 GMT -6
I'm almost always tracking 88.2/24 and mixing analog back to 88.2/32 float, so individual DAC outputs have 24 bit flat dither (post-processing/fader), as well as monitoring output. Clients always have to receive down-converted reference copies, letting DP do the 88.2/32 -> 44.1/16 part and iTunez doing the mp3 (320kbps VBR highest quality) part. I have messed with the LAME framework a little, and so far haven't made an mp3 myself that sounded as 'correct' as what itunez does. 88.2/32 goes to mastering guy, they pick final dither type. If I attend, I can usually hear it, but don't live with any one long enough to ever develop a definite preference. How do you insert the dither plug in post fader? Which Daw are you using....? This is always the part of the equation that never gets a clear answer... cheers Wiz I am not EmRR, however In pro tools you would use a master fader when sending to outboard/ HW... master fader inserts are post fader.
|
|
|
dither
Dec 12, 2016 14:33:27 GMT -6
Post by wiz on Dec 12, 2016 14:33:27 GMT -6
How do you insert the dither plug in post fader? Which Daw are you using....? This is always the part of the equation that never gets a clear answer... cheers Wiz I am not EmRR, however In pro tools you would use a master fader when sending to outboard/ HW... master fader inserts are post fader. I don't use pro tools, so I might misunderstand.. but I was asking about individual tracks... you know, sending 16 tracks for example out to a mixer... how and where he was putting the 16 instances of dither.... I gather a Master Fader is ... well, a Master Fader... unless you can create more than one Master Fader... then.. what is the Master Master Fader... I need a cup of tea.... 8) cheers Wiz
|
|
|
dither
Dec 12, 2016 14:46:43 GMT -6
Post by popmann on Dec 12, 2016 14:46:43 GMT -6
Wow....so, I thought I'd google it for you....fucking Apple. www.logicprohelp.com/forum/viewtopic.php?t=97488There are no post fader inserts. the suggested work around are insanity. That said....I can't imagine you don't have output busses (but I also couldn't imagine not having post fader inserts)....I mean does everything just go to a single master fader, done? My point being, even without the ability to do post fader inserts, you can send 16 channels to output busses, which can have the dither applied.
|
|
|
dither
Dec 12, 2016 14:53:34 GMT -6
Post by wiz on Dec 12, 2016 14:53:34 GMT -6
Wow....so, I thought I'd google it for you....fucking Apple. www.logicprohelp.com/forum/viewtopic.php?t=97488There are no post fader inserts. the suggested work around are insanity. That said....I can't imagine you don't have output busses (but I also couldn't imagine not having post fader inserts)....I mean does everything just go to a single master fader, done? My point being, even without the ability to do post fader inserts, you can send 16 channels to output busses, which can have the dither applied. Thanks Jamie for trying to help out... I worked it out... using the plug ins that Bob mentioned, Bitter (which allows you to view the bit depth) and Airwindows Dither. I was doing some experimenting yesterday and will continue this morning. I was putting Bitter (bit depth plug in) in different parts of the plug in chain within LOGIC and it allows me to see which plugs are making the bit depth go up... its kind of interesting and fun... in a weird way. I shows you how much is going on, which is a lot. I am inserting the dither plug, when needed just before the IO plug in. I am going to do some tests with and without it and see how i fair.. thanks again for trying to help.. cool of you. cheers Wiz
|
|
|
dither
Dec 12, 2016 14:53:58 GMT -6
Post by popmann on Dec 12, 2016 14:53:58 GMT -6
Actually--looking at that longer, are Auxes just their busses you can assign a track to output to instead of the "master fader"? If so, their work around is what I do even though I HAVE post fader inserts....save your tempate with 16 auxes with their dither inserted and name then for the physical outputs--"AnaOut1...AnaOut2....SDPIFOut1, etc"....then you could actually insert the hardware insert IOs there, too. So, you'd go dither in the first slot, then your hardware IO to send it out and return it....then all 16 of those go to the master since for whatever reason you are using an analog mixer but summing digitally. Otherwise, you insert the dither on them then output them to the physical output....stereo input track to recapture the analog sum. So 16 auxes (or however you want to configure mono and stereo)....dither then IO plug in. And my guess? You won't be able to tell the difference. I've never thought it mattered much. But, it's simply not hard to do, and since I DO understand how theoretically it will sound better, sure--I do it.
|
|
|
dither
Dec 12, 2016 15:26:58 GMT -6
Post by Bob Olhsson on Dec 12, 2016 15:26:58 GMT -6
My understanding is that all ASIO is a 24 bit connection.
|
|