Truth and fabulations about recording, rendering, resampling

Discussion of music production, audio, equipment and any related topics, either with or without Ableton Live
Benshik
Posts: 763
Joined: Mon Jan 19, 2004 8:45 am
Location: Moscow/Montreal

Truth and fabulations about recording, rendering, resampling

Post by Benshik » Thu Jun 08, 2006 8:54 am

Hi guys,
long and boring hours ahead: got to bounce every single track of a couple of live sets but I can't chose the best way to proceed. I heard so many rumors about Live's rendering, recording and resampling capabilities that im confused.
Whats the SAFEST way to bounce individual tracks?
(by safe, i mean the one that provides you with the best audio quality and the one that doesnt introduce delays into your recording...)

1- Recording each track to a "recipient" track?
these talks about random delays introduced in the recordings made me paranoid. is really live unable to accurately calculate delays? That process of bouncing is already a real pain, that would be really frustrating to lose even more time re-sync all tracks...

2- Rendering each track separately?
i guess "offline" bouncing doesnt introduce delays, but i heard many people complaining about live bussing system that makes rendered files sound less "good" than resampled or recorded files... Sounds weird to me, but who knows, anyone can develop on this???

3- Resampling each track?
as recording, apparently introduces delays. But this time goes through the master. can it introduce more delays or something?



What do you think?

Pitch Black
Posts: 6444
Joined: Sat Dec 21, 2002 2:18 am
Location: Auckland New Zealand
Contact:

Post by Pitch Black » Thu Jun 08, 2006 11:24 am

I usually use Render To Disk for exporting projects from Live.

Never noticed any problems / artifacts.
MBP i5 2.53GHz | OSX 10.12.6 | Live10.0.4 | Fireface800 | Push 2
https://soundcloud.com/paddyfree

quandry
Posts: 1611
Joined: Tue Mar 18, 2003 2:31 am
Location: San Francisco
Contact:

Post by quandry » Thu Jun 08, 2006 1:52 pm

Pitch Black wrote:I usually use Render To Disk for exporting projects from Live.

Never noticed any problems / artifacts.
second that, no problems here either. Get you mix together, make sure none of the track faders and esp. the master fader are NOT clipping in the red, render to disk (no normalize). Then maybe open up this stereo file in a new set and use a bit of limiting and eq4 to master it and get the output a bit higher. I think there's been a lot of speculation and opinions about this and that, but many people get great mixes all inside of live, not rewiring or rendering each track and using another DAW to render the mix--what a waste of time. If people are that paranoid about Live, why even bother? It works fine for many people, I think these rumors have been mathematically dispelled by Robert Henke as far as I'm concerned.
Dell Studio XPS 8100 Windows 7 64-bit, 10 GB RAM. RME Multiface, Avalon U5 & M5, Distressor, Filter Factory, UC33e, BCR-2000, FCB1010, K-Station, Hr 824 & H120 sub, EZ Bus, V-Drums, DrumKat EZ, basses, guitars, pedals... http://www.ryan-hughes.net

stinky
Posts: 1182
Joined: Mon Jul 25, 2005 1:06 am

Post by stinky » Thu Jun 08, 2006 7:18 pm

I think these rumors have been mathematically dispelled by Robert Henke as far as I'm concerned.
Could you please link to that post, or any other link that you used as a reference for your quote? Thank you.

quandry
Posts: 1611
Joined: Tue Mar 18, 2003 2:31 am
Location: San Francisco
Contact:

Post by quandry » Thu Jun 08, 2006 7:33 pm

stinky wrote:
I think these rumors have been mathematically dispelled by Robert Henke as far as I'm concerned.
Could you please link to that post, or any other link that you used as a reference for your quote? Thank you.
sure, here:

http://www.ableton.com/forum/viewtopic. ... highlight=

note the first post as "guest" is actually Robert Henke, this was before the days of a bunch of asswipes mucking up the forum with garbage, before the days of manditory logins (which, since they are anonymous, haven't really cut down on the crap).
Dell Studio XPS 8100 Windows 7 64-bit, 10 GB RAM. RME Multiface, Avalon U5 & M5, Distressor, Filter Factory, UC33e, BCR-2000, FCB1010, K-Station, Hr 824 & H120 sub, EZ Bus, V-Drums, DrumKat EZ, basses, guitars, pedals... http://www.ryan-hughes.net

stinky
Posts: 1182
Joined: Mon Jul 25, 2005 1:06 am

Post by stinky » Thu Jun 08, 2006 7:58 pm

I appreciate that info greatly, but i'm still not completely convinced. First, here's another statement by Henke that i choose to quote regarding sound quality:

http://www.ableton.com/forum/viewtopic. ... 8&start=15

I just think there are too many variables to factor in to make such a broad statement equate. Yes, it's up to the individual producer, a/d, equipement, etc. But, i don't think this statement "Live sounds like any other DAW" should be taken literally, and that he categorically proved anything in that post. He only "proved" that any errors live makes in conversion (if any) are too small to distinguish. But saying "Live sounds like any other DAW" is the same as saying a "BMW drives like any other Car", which just isn't true.

Regardless, for the original poster, to save you time, i think you should take a look at this post:
This should save you alot of time, and if you're worried about artifacts, just keep track of your warping (and test out different warp modes, if you need to) before you export.

quandry
Posts: 1611
Joined: Tue Mar 18, 2003 2:31 am
Location: San Francisco
Contact:

Post by quandry » Thu Jun 08, 2006 8:17 pm

stinky wrote:But saying "Live sounds like any other DAW" is the same as saying a "BMW drives like any other Car", which just isn't true.
thats quite a bit of an overstatement there. Software boils down to 1's and 0's, and the general algorhythms and summing procedures are pretty similar in theory amongst various DAWs. I guess there could be differences, but there is not way on God's green earth that the differences are as fundamentally different as a BMW vs. my p.o.s. subaru wagon from 1990. There are so many variables to cars, on a scale of 1-10 my subaru is like 1.5, and a beamer is 10. On an audio production scale of 1-10, if you think PT is 10, Live is at least 8+, while a radio shack tape recorder=my car. Anyhow, Live works for me...

Someone should use the trick in your second link as a test: record all of your tracks in Live (with effects) into new clips to bring into PT. Then render the mix in Live with no master channel effects, then bring these individual tracks into PT, leave everything set at zero, no extra effects or anything, and render it there with the same master volume level as you had in Live (which I'm assuming is less than zero or else clip city). Then listen to the two tracks, A-Bing between them, and see what (if any) difference there is. Then post them for others to listen to.
Dell Studio XPS 8100 Windows 7 64-bit, 10 GB RAM. RME Multiface, Avalon U5 & M5, Distressor, Filter Factory, UC33e, BCR-2000, FCB1010, K-Station, Hr 824 & H120 sub, EZ Bus, V-Drums, DrumKat EZ, basses, guitars, pedals... http://www.ryan-hughes.net

stinky
Posts: 1182
Joined: Mon Jul 25, 2005 1:06 am

Post by stinky » Thu Jun 08, 2006 8:55 pm

thats quite a bit of an overstatement there. Software boils down to 1's and 0's, and the general algorhythms and summing procedures are pretty similar in theory amongst various DAWs. I guess there could be differences, but there is not way on God's green earth that the differences are as fundamentally different as a BMW vs. my p.o.s. subaru wagon from 1990.
I call bullshit!! There's no way that "general algorythms" and "summing procedures" in "theory" means that it will be implemented that way in practice. That's just a rediculous statement on your part, i'm sorry. You're not taking into account that humans code this shit, and everyone has a different way of doing things. Just because the procedures for making plastics and metals generally have been the same, doesn't mean that they're the same quality from each manufacturer. That's why you have brand names, and rip offs. The general "algorythms" and "procedures" for making cheese, beer, wine, etc have been established for centuries, but as we all know not all taste the same. Even if you used the same ingredients, the same processes and procedures from two different manufacturers, it would come out differently.. why, because of human comprehension, perspective, and ultimately, error. There are 100s of ways to implement an FFT.. When corn flakes first came out, there were 100s of manufacturers, and who do you think if the most dominant now? Kellogs, and why? Because noone could reproduce their freakin cornflakes, and everyone tried.

Sorry, that's just naive to think to that way.

As for the test, those are definitely good criterion to go by... if I had time, i would do it (and just might do that for this test), and hope that all would trust me to be fair and impartial.. I have done some testing in that regards, and found the sound i got from PT was the best... but, i didn't test using stringent guidelines, and would be very interested to know the truth, and If i'm wrong, i'll be very forthright.

I've outlined my methods in other posts, and why i choose to rewire to PT.. if i'm doing that for no reason, and just wasting my time, i'd definitely rather know. I've discussed some very specific reason for why i do what i do here:
I'm not here to flame, but there has been no conclusive evidence to support either claims, yours or mine... only assumptions, and it would definitely be great to put this issue to bed once and for all.

If any of the other PT users out there can help out with this test, that would be great!

Angstrom
Posts: 14678
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Post by Angstrom » Thu Jun 08, 2006 9:21 pm

stinky wrote: I'm not here to flame, but there has been no conclusive evidence to support either claims, yours or mine... only assumptions, and it would definitely be great to put this issue to bed once and for all.
never going to happen though is it.

because unless someone here can hear 32bit rounding or truncation it's impossible to say exactly whats going on. The only way we would gain an insight is for Gerhard to post "oh I just noticed my bussing code is actually 16 bit, oops, I better fix that! "
I can't see that happening somehow.
The closest we will get is marketing blurb like - "Live 7 with new improved sound engine", that's when you will know the truth. ;)

jerry123
Posts: 300
Joined: Wed Oct 12, 2005 3:26 am
Location: Toronto Ont. Canada

Post by jerry123 » Thu Jun 08, 2006 9:27 pm


Angstrom
Posts: 14678
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Post by Angstrom » Thu Jun 08, 2006 9:32 pm

jerry123 wrote:Check this out.
http://src.infinitewave.ca/
I've not seen the point of Ableton in that test, as it's a test of converting a 96khz wave to 44.1 , nothing else as far as I can tell.

is that something that any of you do in Live?
Resampling numerically unrelated samplerates is something you do in a high end wave editor surely.

jerry123
Posts: 300
Joined: Wed Oct 12, 2005 3:26 am
Location: Toronto Ont. Canada

Post by jerry123 » Thu Jun 08, 2006 9:34 pm

The test basically gives an example of the quality of conversion that takes place within the software. You can use the drop down list to compare Live to other pieces of software.

Benshik
Posts: 763
Joined: Mon Jan 19, 2004 8:45 am
Location: Moscow/Montreal

Post by Benshik » Fri Jun 09, 2006 6:28 am

stinky wrote: Regardless, for the original poster, to save you time, i think you should take a look at this post:
hey stinky, thanx for the suggestion, that indeed can be a time saver...

so is that method as safe as rendering??? can it introduce delays? if yes, then id rather waste some time on rendering tracks one by one.

quandry
Posts: 1611
Joined: Tue Mar 18, 2003 2:31 am
Location: San Francisco
Contact:

Post by quandry » Fri Jun 09, 2006 2:19 pm

Benshik wrote:
stinky wrote: Regardless, for the original poster, to save you time, i think you should take a look at this post:
hey stinky, thanx for the suggestion, that indeed can be a time saver...

so is that method as safe as rendering??? can it introduce delays? if yes, then id rather waste some time on rendering tracks one by one.
hmm, I guess with various reverbs and vst you could get slight delays due to the time it takes to run through effects, but technically plug-in delay compensation should take care of it. But I think you are right, the overall safest bet would be to have no master effects, set the master fader to zero, and render each track to disk.
Dell Studio XPS 8100 Windows 7 64-bit, 10 GB RAM. RME Multiface, Avalon U5 & M5, Distressor, Filter Factory, UC33e, BCR-2000, FCB1010, K-Station, Hr 824 & H120 sub, EZ Bus, V-Drums, DrumKat EZ, basses, guitars, pedals... http://www.ryan-hughes.net

quandry
Posts: 1611
Joined: Tue Mar 18, 2003 2:31 am
Location: San Francisco
Contact:

Post by quandry » Fri Jun 09, 2006 2:30 pm

stinky wrote:
thats quite a bit of an overstatement there. Software boils down to 1's and 0's, and the general algorhythms and summing procedures are pretty similar in theory amongst various DAWs. I guess there could be differences, but there is not way on God's green earth that the differences are as fundamentally different as a BMW vs. my p.o.s. subaru wagon from 1990.
I call bullshit!! There's no way that "general algorythms" and "summing procedures" in "theory" means that it will be implemented that way in practice. That's just a rediculous statement on your part, i'm sorry. You're not taking into account that humans code this shit, and everyone has a different way of doing things. Just because the procedures for making plastics and metals generally have been the same, doesn't mean that they're the same quality from each manufacturer. That's why you have brand names, and rip offs. The general "algorythms" and "procedures" for making cheese, beer, wine, etc have been established for centuries, but as we all know not all taste the same. Even if you used the same ingredients, the same processes and procedures from two different manufacturers, it would come out differently.. why, because of human comprehension, perspective, and ultimately, error. There are 100s of ways to implement an FFT.. When corn flakes first came out, there were 100s of manufacturers, and who do you think if the most dominant now? Kellogs, and why? Because noone could reproduce their freakin cornflakes, and everyone tried.
I'm not trying to flame either, and sounds like you may know more about the technical aspects of software programming than me. But, do you really think your analogies are that fair--c'mon the range of qualities in cars, beers, cheese, wine, etc. varies hugely from product to product--like the BMW vs. my 1990 subaru, or thunderbird vs. crystal. I still stick my statement that if the pinnacle of digital procution sound is PT or Logic or whatever at a perfect "10", in my book Live can't be less than an "8"--the difference isn't really that drastic--a crappy $15 tape recorder would be a 1. Whereas the difference between cars, cheese, beer, and wine spans the scale from 0-10.

Maybe somebody will do a test, or Robert can chime back in. I'm still inclined to believe his statement in the post I linked to above--he certainly knows way more about music software, Live, and programming than me--I tend to take his word for it since he is so involved. Non offense, but do you think you know more than him about these matters? From his statement, it does imply/state that warping clips, or rendering to 16/44.1 from a higher bit or sample rate could resulting in differences between various apps, which seems to be supported by jerry 123's link.
Dell Studio XPS 8100 Windows 7 64-bit, 10 GB RAM. RME Multiface, Avalon U5 & M5, Distressor, Filter Factory, UC33e, BCR-2000, FCB1010, K-Station, Hr 824 & H120 sub, EZ Bus, V-Drums, DrumKat EZ, basses, guitars, pedals... http://www.ryan-hughes.net

Post Reply