|
|
Post by tonycamphd on Jun 14, 2015 17:18:51 GMT -6
Ok, i picked up another keyboard, so heres some good reading on digital filters, and here's a thread started by one our favorite RGO guys, with some serious..., and I mean SERIOUS heavy weights chiming in on the subject(Sir Bob O included) repforums.prosoundweb.com/index.php/topic,5739.0.html (apparently you have to copy this entire link and paste it in your address bar, for some reason when you click it, it leaves off the tail end?) www.users.qwest.net/~volt42/cadenzarecording/Filters.pdfhere's slightly "smug" Monty doing a great job of explaining things until he says "no one ever ruined a great recording by not employing dither", he fails to realize that while mixing, some people use upwards of 100 tracks with a zillion plugs that have quantization noise all over them, the signal dependent, nasty upward harmonic noise build up, and subsequent cascaded passing over one another build up to VERY audible artifacting levels.. aka "swirling space monkey digititus", I believe this is what bothers me so about digital recordings. Or, I'm totally wrong, and it's something else entirely haha ahhh yes the thread from my past... 8) when a young (well younger) bright eyed and bushy tailed wiz, wandered into the electronic lounge room of some of the greatest names in engineering history and asked a question that was really in the context of the company, quite a basic and inane question, and was very graciously put in his place by the inventor of the parametric EQ. That thread, is what is great about the internet... Bob , George, Nika and others... all very brilliant and extremely talented guys, taking their time to explain to a nobody like me... How f*&Cking cool is that.... how gracious people can be. The internet was once a place (and RGO might be the last vestige of this) where if you were well mannered and respectful, people like this, would take time out of their day, to explain to people like me... Honestly, this thread, and the couple of phone calls I had with Bob, are right up there for me. cheers Wiz PS how time flies eh? you handled yourself superbly by any standard amongst those giants buddy! I'd have pissed my pants like a scared schoolgirl lol |
|
|
|
Post by jazznoise on Jun 14, 2015 17:48:11 GMT -6
I have a thread regarding my final mix sounding different than my listening from my DAW. Could it beg that when I bounce ( I use Logic X), I'm doing something wrong? Although I have some tracks don at 44.1, I use 48k now always. I'm really only concerned with my 24 bit WAV files staying 24 bits for the final mixdown. I'm not sure I'll ever use the CD format again, so I don't know if I'll need 16 bits for any purpose now. I will need a good sounding mp3 copy though. It could be. I know with my laptop it only uses my chosen soundcard device for playback in DAW's and Google Chrome, so using VLC or Windows Media Player or iTunes always sounds different because the Realtek HD drivers f*ck up the sound with some EQ/Compression crud. Try reimporting your final mix into your DAW and see if that fixes it. The sound of quantization distortion is a tricky topic - I guess I'd compare it to Crossover Distortion in that it's more apparent on quiet signals and it's mostly just inharmonic junk relating to the bottom bit repetitively swinging from 0 to 1. Going from 32-24 isn't as serious a dither crime because the quantization distortion is still sitting at roughly -140dB. However if it's not done it could funk up things later when further downsampling. Either way, this should be a question for DSP engineers - in the same way that I shouldn't have to worry about how the Sinc function for an ADC is implemented. |
|
|
|
Post by swurveman on Jun 14, 2015 18:13:20 GMT -6
Correct me if I'm wrong, but here's the most definitive statement I got from the linked rep forum discussion:
"If the bit depth is being reduced then you dither - period. So if you're going internally from a 32 bit numerical nomenclature to 24 bits on output you dither - period. If you are going from 24 bits to 16 bits you dither - period. If you are going from 48 bit to 24 bit you dither - period."
Here's the question I have: During a mix, does this just apply only to the final bit depth reduction on the summed Mix Bus "Bounce to Disk" (PT) or "Audio Mixdown" (Cubase), or should this also be happening at the individual channel level when you'r mixing hybrid and have a hardware insert on a channel and you're mixing in a 32 bit float DAW but your A/D converter that is processing the output and return to the hardware is 24 bit? In other words, should I have a 24 bit dither plugin on the channel that has the hardware insert?
|
|
|
|
Post by b1 on Jun 14, 2015 18:26:07 GMT -6
From what I understand in regards to Media Players, they all handle the audio differently, which is why a file will sound different in diff Media Players. I recall developers discussing this to great length
and mentioned that JRiver Media Player handled rendering/playing the file the best; along with ASIO drivers.
|
|
|
|
Post by delcampo on Jun 14, 2015 18:38:49 GMT -6
In an attempt to simplify (a thus far elusive pursuit) ...be very succinct, & as I understand it from others that admonish to significance of dither... - Final mix from 32 float to 24 = Dither of course. - ANY channel that feeds a D/A (to feed a console or HW etc.) = Dither (especially if the channel has ANY DSP on it) - For any internal channel "bounce" to another internal DAW channel = Dither (especially if the channel has ANY DSP on it) Folks more informed than I, (ie; almost everyone  ., please feel free to point out anything that is not correct in these scenarios or, If there is another situation in which one should dither. |
|
|
|
Post by mulmany on Jun 14, 2015 19:04:43 GMT -6
Correct me if I'm wrong, but here's the most definitive statement I got from the linked rep forum discussion: "If the bit depth is being reduced then you dither - period. So if you're going internally from a 32 bit numerical nomenclature to 24 bits on output you dither - period. If you are going from 24 bits to 16 bits you dither - period. If you are going from 48 bit to 24 bit you dither - period." Here's the question I have: During a mix, does this just apply only to the final bit depth reduction on the summed Mix Bus "Bounce to Disk" (PT) or "Audio Mixdown" (Cubase), or should this also be happening at the individual channel level when you'r mixing hybrid and have a hardware insert on a channel and you're mixing in a 32 bit float DAW but your A/D converter that is processing the output and return to the hardware is 24 bit? In other words, should I have a 24 bit dither plugin on the channel that has the hardware insert? The answer is yes. If you use hardware inserts or process out of the box you dither. I have been experimenting with dithering on individual tracks even if I am not going OTB. I have found that it smooths out the sound of the tracks, no digititus as Tony puts it. Finding this really true when placed before an Eq with lots of curves in it, or a track with lots of volume automation. Thanks to everyone for bringing up this topic and forcing me to really listen and experiment with different placement of dither plugs. |
|
|
|
Post by jazznoise on Jun 14, 2015 19:20:37 GMT -6
From what I understand in regards to Media Players, they all handle the audio differently, which is why a file will sound different in diff Media Players. I recall developers discussing this to great length and mentioned that JRiver Media Player handled rendering/playing the file the best; along with ASIO drivers. Well the first big issue it depends on is the type of file. Compressed files have codecs, and the proper implementation of the decompression algo in the codec will probably dominate issues there, problems made here get worse down the line. Wav is just read as it is - it's an interleaved stereo file and so you pass each word to its appropriate channel. Stuff like VLC does this cleanly with no junk. But after your program choice you still have another software layer or 2 before the hardware layer. If your media player allows you to chose what driver you use, then always use something like ASIO that just acts like a "wire" with a buffer. However laptops tend to have built in drivers that always get called, and they often have particular pre-EQ and pre-Comp stuff to counteract the volume and bandwidth issues of the speakers - even on the Headphone outs! |
|
|
|
Post by Martin John Butler on Jun 14, 2015 19:26:15 GMT -6
I use Logic X. When I've mixed a track, and then bounce, what settings do I need to use? When I read "So if you're going internally from a 32 bit numerical nomenclature", I don't know, but would guess that Logic is running at 32 bit ?
|
|
|
|
Post by b1 on Jun 14, 2015 19:49:41 GMT -6
From what I understand in regards to Media Players, they all handle the audio differently, which is why a file will sound different in diff Media Players. I recall developers discussing this to great length and mentioned that JRiver Media Player handled rendering/playing the file the best; along with ASIO drivers. Well the first big issue it depends on is the type of file. Compressed files have codecs, and the proper implementation of the decompression algo in the codec will probably dominate issues there, problems made here get worse down the line. Wav is just read as it is - it's an interleaved stereo file and so you pass each word to its appropriate channel. Stuff like VLC does this cleanly with no junk. But after your program choice you still have another software layer or 2 before the hardware layer. If your media player allows you to chose what driver you use, then always use something like ASIO that just acts like a "wire" with a buffer. However laptops tend to have built in drivers that always get called, and they often have particular pre-EQ and pre-Comp stuff to counteract the volume and bandwidth issues of the speakers - even on the Headphone outs! What you mentioned jogged my memory, so it's good to note the discussion I referred to was predominately regarding wav files - before VLC and Win7 existed... I'm using VLC as default player here. |
|
|
|
Post by Bob Olhsson on Jun 14, 2015 22:37:17 GMT -6
Basically you have a choice between distortion and d*ther every single time you reduce bit depth. Most people fail to realize that the d*ther prevents distortion as opposed to just covering it up.
You can get away with a succession of floating point calculations because it pushes the level all the way up before truncating the bottom but it's a case of getting away with it as opposed to doing mathematically correct DSP.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 15, 2015 2:23:51 GMT -6
For the question about DAW internal formats (no matter if Logic or whatever modern DAW):
Modern DAWs AFAIK all use floating point formats internally. 32bit or 64bit floating point is ALWAYS higher bit depth i.e. represents audio in a higher resolution, than the PCM formats sent to file or DAC. These use fixed point formats like unsigned integer to represent audio. So there is downwards quantization happening. So: dither recommended.
32 bit float is higher resolution than 32 bit unsigned integer format (32 bit standard wav file).
So beware that we are talking different things if talking about 32bit and 64bit modes of internal DAW processing (float!) and bit depth of the soundcard drivers and files use.
Apples and oranges...
|
|
|
|
Post by Johnkenn on Jun 18, 2015 9:24:56 GMT -6
Just wanted to say - I applied dither to all of the channels goin out to my outboard and coming back in...and I swear I hear more clarity than the results in the past. It could be a placebo, but who cares...I think I'm hearing better results, so I'm gonna start implementing it.
|
|
|
|
Post by tonycamphd on Jun 18, 2015 10:23:19 GMT -6
Just wanted to say - I applied dither to all of the channels goin out to my outboard and coming back in...and I swear I hear more clarity than the results in the past. It could be a placebo, but who cares...I think I'm hearing better results, so I'm gonna start implementing it. if your DA/ADing all those tracks like i believe you are? It's not a placebo effect. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 18, 2015 10:41:13 GMT -6
John, dither before going out to outboard via DAC is totally correct, but on the ins? You mean you apply dither to incoming AD channels? This should only increase the noise level...?
|
|
|
|
Post by Bob Olhsson on Jun 18, 2015 13:34:01 GMT -6
Coming in the d*ther needs to be in the output of the A to D converter. This is set with jumpers if settings are available at all and generally is not a user option.
|
|
ericn
Temp
Balance Engineer
Posts: 16,107 
|
Post by ericn on Jun 18, 2015 15:20:46 GMT -6
Here is how I found I liked dither, CD's always sounded better with the sun roof open and all the noise !
The Dither was a must after that !
|
|
|
|
Post by Johnkenn on Jun 18, 2015 16:44:53 GMT -6
John, dither before going out to outboard via DAC is totally correct, but on the ins? You mean you apply dither to incoming AD channels? This should only increase the noise level...? No. Just dithering going out. Who said anything about in? |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 18, 2015 19:00:40 GMT -6
John, dither before going out to outboard via DAC is totally correct, but on the ins? You mean you apply dither to incoming AD channels? This should only increase the noise level...? No. Just dithering going out. Who said anything about in? Oh, sorry John, read it again and it is a misunderstanding of mine, i guess due to the fact english is not my first language...! |
|
|
|
Post by Bob Olhsson on Jun 19, 2015 10:04:17 GMT -6
Don't feel bad, I missed it too!
|
|
., please feel free to point out anything that is not correct in these scenarios or, If there is another situation in which one should dither.
