rendering question

[tw1nstates]

Someone raised an interesting point in the Analogue thread.

Running your project at a higher sample rate wll make some of the ableont fx sound better.

ok cool.

What about rendering?

ie. i have got a project that I can't turn to 192khz but if I render at 192 then does it in turn calculate all of the audio processes at 192 - thus saving me from going back and bouncing everything (again) to test this. . .

If anyone knows that would be excellent.

Just to clarify i have a project running at 44.1 I want to render at a high sample rate *if* it makes a difference to he sound of the internal plugs - I have a few things which theoretically should sound better at a high sample rate.

be interested to know if anyone knows. . .

Thanks!

[roothewhirl]

I do know that most cds are sampled down into 44.1 at 16 bits anyways. This is part of the debate when it comes to the best rendering mode. I also know that mastering engineers do prefer the highest bit rate and sample rate possible.

I'm not sure if there is a massive difference in the dither of say a reverb tail at 44.1 vs 192, (unless you crank the gain of the tail, which is a bit impractical.) But unless someone is listening to it on some crazy hi fi system (what? 1% of music listeners?) or it's being played through a space age club system, 44.1 should be fine. Could be wrong, though. Good luck!

-Ryan

[leedsquietman]

Dan Lavry, who makes the high end studio Lavry converters (i.e. Lavry Golds) maintains that using sample rates above 96 Khz is not worth it and wrote a paper saying how 192 Khz sounds worse. http://www.lavryengineering.com/product ... e4496.html



It might help with some of your plugins to render at higher samplerate, which can help prevent aliasing if you have a lot of FX processing going on, but for the most part, rendering at 44.1 or 48 is perfectly adequate, especially for pop/rock/electronic type music. Classical music and jazz aficionados sometimes request 24/96 recordings.

Your audio itself won't sound better if you recorded it at 24/44.1 and you upsample it to 88.2 or 96 Khz. It's only plugins which can gain. Of course, you may have recorded your audio at 88.2 or 96 Khz in the first place, which to my ears makes a very tiny, subtle difference on acoustic instruments, or brass or stringed instruments in a sparse mix, but we are talking a very minor difference, which comes with a huge trade off in terms of CPU load and reduced track and plugin counts.

[elxicano]

I'd personally agree that it is pointless above 96, which is already considered High Definition audio.

For processing your 44.1 audio through plugins, YES, there is *potential* for having a better sound quality, rending at a higher rate.

Still, what's better is if you are working at least 48K (which would be pointless if your audio source is less). Consider it headroom for mangling and editing without losing much quality in sound. This can be important if you want to release in CD quality.

The thing to keep in mind is that every permanent change you make to the audio file has potential for sound degradation. In programs like Live and most DAW's, this is a minimal risk since adding EQ, Compression, Limiting, etc... won't change the actual file until you render it, but it's more important in dedicated Audio Editing programs like Wavelab and SoundForge, which make hard changes to the audio source.

Because of what I mentioned above, it is okay to work in 44.1K without causing permanent degradation to the audio, but if you decide to render at a higher rate, then make sure that after changing the audio rate, close and reopen your project before rendering, since it can sound different after making the change AND some plugins don't work properly after changing the rate (such as Kontakt to name one). I strongly suggest to not simply render at a higher rate, but instead change the rate first... close, then reopen your project and THEN render.

[Tone Deft]

Someone raised an interesting point in the Analogue thread.

Running your project at a higher sample rate will make some of the ableont fx sound better

that point is wrong. 3phase wrote it, that should be warning enough.

don't believe everything you read in that thread.

[tw1nstates]

Thnaks for the replies.



Think I am gonna have to call the ableton support people for their take on it.

I know that upsampling occurs a fair bit in some plugins and that running at a higher sample rate that 44.1 is generlly considered better (and that it's debatable as to whether running at higher than 96 is going to give you better results - didn't know that it would perhaps give you worse results though).

Tone, why do you think that rendering so that you run your plugs would give a worse or same result? Am curious to know.

Also, the important point here is that it's Audio unit synths and plugs for the most part in the project (some field recordings and loads of sampled percussion) - so, I should get a difference - running NI stuff at higher sample rates makes it sound for the most part smoother. . .

But, does Ableton render the internal processes at 96 or does it render them at 44.1 and give you a 96 final file (comprised of loads of 44.1 files summed together in the mix bus)

Thanks

[elxicano]

But, does Ableton render the internal processes at 96 or does it render them at 44.1 and give you a 96 final file (comprised of loads of 44.1 files summed together in the mix bus)

Thanks

If you choose 96, then it renders at 96, but of course if you are working with WAV files recorded at 44.1, then depending on what audio processing plugs you've chosen, there may be absolutely no difference in quality or sound.

[Tone Deft]

Tone, why do you think that rendering so that you run your plugs would give a worse or same result? Am curious to know.

I have ears and can think for myself. I've used Live for 6 years and have never noticed a difference when I change sample rates. 3phase says that if you switch Live from 48k to 96k then the sound difference is obvious, I don't mean plug-ins I mean the program itself. try that one out for yourself and see if he's full of shit.

3phase pops up in here and rants about some serious bullshit. that's fine, until people start reading what he posts and taking him at his word. the guy's a complete bullshitter. he name drops pieces of gear to support his cred but if you actually read his theories it's pretty obvious the guy's a bullshit artist.

you own Live, try 3phase's theories for yourself. does Live sound noticeably different at 48k vs 96k? do the plug ins sound different at different sample rates?

not to mention, what are your monitors? what are your pre-amps? what tools do you use to record? are ALL your other recording and mastering techniques well practiced, understood and up to snuff to the point where your mastering is so outstanding that getting rid of artifacts at -100dB is worth the time? it sounds cool and deep to say you can hear the difference between sample rates but it's all bullshit posturing.


thanks for keeping an open mind.

oh, and throughout all this, I'd rather record a great performance than polish the turd of a bad performance. some of my favorite albums sound like shit. early New Order, Depeche Mode, Cure, Smiths, most of Hendrix's recordings, all sounds like crap. just an irony I like to keep in mind, KEEP IT SIMPLE.

[Harris.Andrew]

Someone raised an interesting point in the Analogue thread.

Running your project at a higher sample rate wll make some of the ableont fx sound better.

ok cool.

What about rendering?

I think the point was that warping (and possibly some other effects, but definitely some modes of warping) should behave a little better with higher sample rates. The critical theory here is Nyquist sampling frequency - this gets stretched as well when audio is sped up or slowed down, and higher sampling rates may allow for information which may not have been necessary at one speed to sort of settle into audible frequencies after stretching.

I've played with this a bit in the past and was pretty sure higher sampling rates did indeed lead to better stretching for some warp algos, or better behavior when sampled. But it wasn't perfection or anything. Also, the warping modes have changed a little since then - the transient mode on beats warping is great for percussion loops and the complex/pro modes are pretty stellar for e.g. DJ mixing. Wouldn't swear on any of this though.

For rendering a full set, I don't think there's much of a reason to go with huge sample rates.

[Tone Deft]

The critical theory here is Nyquist sampling frequency - this gets stretched as well when audio is sped up or slowed down, and higher sampling rates may allow for information which may not have been necessary at one speed to sort of settle into audible frequencies after stretching.

I disagree with that speculation. have you actually noticed a difference or have you just talked yourself into thinking this makes sense so it must be true?

the sampling theorem states that if you sample at more than twice the frequency is the signal in question then you can accurately reproduce that signal. the notion that a higher sample rate has more information in it so warping works better does not make sense to me.

it's pure speculation. I'm very much open to listening tests and hearing from someone that knows how the warping modes actually work but the crux of your statement doesn't fly with me.

IMO. it really is all opinion. I might say it's up to you to provide proof of this theory's existence because one doesn't have to support the non-existence of something.

[elxicano]

The critical theory here is Nyquist sampling frequency - this gets stretched as well when audio is sped up or slowed down, and higher sampling rates may allow for information which may not have been necessary at one speed to sort of settle into audible frequencies after stretching.

I disagree with that speculation.

the sampling theorem states that if you sample at more than twice the frequency is the signal in question then you can accurately reproduce that signal. the notion that a higher sample rate has more information in it so warping works better does not make sense to me.

it's pure speculation. I'm very much open to listening tests and hearing from someone that knows how the warping modes actually work but the crux of your statement doesn't fly with me.

IMO. it really is all opinion. I might say it's up to you to provide proof of this theory's existence because one doesn't have to support the non-existence of something.

Not exactly. See page 98 of your manual in reference to Hi-Q mode... As I've been saying before, it makes absolutely no difference if you are working with audio set at 44.1k and then rendering at 96 (unless the effects are sampled at higher rate, but speaking specifically about warping now). However, if you have a sample that was recorded at 96, and you want to use warping, you will have less "distortion" if you either switch your sample rate to 96 OR enable Hi-Q mode (which uses more CPU).

It's probable that whomever said you'd have a higher quality using warping at higher sample rate, was working with sample already recorded at a higher sample rate.

[Tone Deft]

Hi-Q mode should always be on.

ethios4 posted a nice plot of the difference of Live's SRCs in Hi-Q vs without it. someone at Ableton even chimed in saying that Live's SRC is not its strong point and if you have any SRC questions use another software program.

nice post, thank you for facts.

[elxicano]

Hi-Q mode should always be on.

ethios4 posted a nice plot of the difference of Live's SRCs in Hi-Q vs without it. someone at Ableton even chimed in saying that Live's SRC is not its strong point and if you have any SRC questions use another software program.

nice post, thank you for facts.

Thanks for adding Hendrix to your list. IMO, those crappy recordings are absolutely beautiful.

BTW... I don't use warping, so all I have to comment on is the manual.

[leedsquietman]

DAWS typically don't have great SRC, including some of the more trad versions.

Problem is solved by using Ozone's SRC or Voxengo's R8Brain (even the free version). I never SRC in any DAW.

Audio recorded at lower rates but upsampled is not going to improved.

[Cervantes]

great thread. i'm going into mastering on monday after an album i've been working on for a few years, all in live 6 =) - yeah can't afford upgrade just yet.

can someone recommend what to do here. most of my project sets contain a mixture of 44.1/16 bit and 48k/24 bit. what should i render to? or should i convert all the 48k samples to 44.1 before rendering? is it even worth doing this?

final output are mp3's initially for a free online giveaway relase, and then to CD.

thanks, and aps if i've missed the answer above, was getting a little techie for me =)