Any Improvements In The Audio Engine Of Ableton Live 10?

[jestermgee]

When this kind of topic arises, I always feel like "seriously, don't you have anything else to care about?".
We're in 2017; if you can't make your music sounding good with what's available maybe you should blame someone else than your tool(s).
Even if we'd suppose that Live's "engine" doesn't good as other daw's ones, just for the sake of supposing, then just use you equalisers, saturations, reverbs and compressor to fill the gaps. Just adapt your mixing to your daw's "inferiority". The same way you would do to counteract your mixing room's imperfections.

+1

How many more times does this topic need to arise, with no actual evidence scientific evidence, no outcome and no agreement for/against the argument. Every piece of "test data" I see is never done in a proper controlled fashion with the project files offered up for dissection and to look at the results (or listen to them) if there was any difference, it's so unnoticeable it's difficult to understand how that could affect your game.

Wish I had that much time to ponder...

[Martin Gifford]

There's a few reasons I think Live has gotten a bad rap about this over the years:

- Warping tends to be on be default, and if there's any tempo differences that will trigger timestretching which obviously alters the sound.
- Fades on clip edges is on by default, which can affect transients and help lose a sense of depth in some cases.
- Different DAWs use different panning laws, some are just stereo balance instead of true panning too (thankfully fixed in Live 10).
- SRC. In the past Live's sample-rate conversion was so so. Fine for live use, but not always the greatest for the studio. Ableton themselves in the past recommended doing any SRC in another app before bringing the files into Live. This was greatly fixed in Live 9 with the addition of the SOX algorithm, though this is only used when downsampling and not upsampling.

We should actually make a list. To the above, I'd add:

- The old problem with delayed processing of effects, or something like that, which has been fixed now.
- Dull Ableton instruments and effects, which is fixed by using third party plug-ins, but the new Wavetable synth seems good.

There's also the psychological and subjective issues like the previous dull GUI might have made people feel depressed. Also, using the Intro version of Ableton might make people think Ableton sounds bad because the Intro version has some truly ugly sounding "instruments" that cause pain to your musical soul.

[Stromkraft]

i think you are over-simplifying things. there is no single test that suffices. even if you play the same audio file in two different DAWs, you cannot assume that each audio engine outputs the audio exactly the same.

in the end is unlikely that basic digital summing that is the most audible issue, if it is even an issue at all. if we are talking about mixes sounding different, when then there are whole range of other possible differences, many of which are explored in Tarekith's experiments.

double-blind testing is an important method because it shows whether or not we can reliably perceive a difference over a number of trials, and it illuminates how subjective our perceptions are.

I disagree with your premise. It was stated that it is the digital summing that is causing audible differences is sound quality.

How do you mean that 2 work products from two DAWs, that do null more or less completely, can indicate anything else than they sum the given tracks exactly the same? And if they do there clearly is no difference in the summing area. That's what I think anyway.

It's endlessly more complicated to make a judgement call on what is preferred. Which is why I didn't enter into that discussion yet. But for the record A/B testing is also flawed. It's biased to what's consciously familiar. This doesn't mean those tests are without value in some areas. I don't see this apply well here though. Worth a try I suppose, but they tests better be rigid.

But for mixing tools I'm a strong believer in that you should choose those that serve you well. Not everyone is skilled in this and some even lack the skill to know when they need to ask for the service of other people, but that's how it is. What would the alternative be than to make your choices and live with them?

[tottigoool]

Hi,
I've another question about the Audio Engine in Live 10. Live 9 had some PDC / ADC issues that has been solved. But there are still some elements that are still not compensated like the "level meters". Mixing with a high number of UAD plugin can be very exhausting. Another important function is "Reduced latency when monitoring" which seems to work only if plugins with higher latency are not in the same track as your VST Instrument. The problem is that you can't play VSTi without extra latency as long the chain or routing includes plugins from UAD for example. To make things short: Has Live 10 a better "Low-Latency-Mode" which disables or deacitvates temporarily all plugins that add latency to the audio path?
Thanks...

[2pauluzz2]

The old problem with delayed processing of effects, or something like that, which has been fixed now.

Do you mean this?

Ableton calculates delays from left to right. Getting to the sends last. If you then route the audio from the sends back into the project they will forever be late. And that's where you get the delay from.
I stopped using sends in Live for this very reason. Instead I use racks with 2 channels one wet and one dry and blend it using the channel volumes.
It's not really a bug. But rather a limitation in the way Live implemented PDC. The thing is that they allow for complex routing , but once you become a power user you quickly find these kind of limitations, and it sucks.

If yes, is it fixed in 10? Because it's still an issue in 9 as far as I know.

[2pauluzz2]

Posting at the same time about the same topic.. coincidence.

[Stromkraft]

The old problem with delayed processing of effects, or something like that, which has been fixed now.

Do you mean this?

Ableton calculates delays from left to right. Getting to the sends last. If you then route the audio from the sends back into the project they will forever be late. And that's where you get the delay from.
I stopped using sends in Live for this very reason. Instead I use racks with 2 channels one wet and one dry and blend it using the channel volumes.
It's not really a bug. But rather a limitation in the way Live implemented PDC. The thing is that they allow for complex routing , but once you become a power user you quickly find these kind of limitations, and it sucks.

If yes, is it fixed in 10? Because it's still an issue in 9 as far as I know.

What do you mean fixed? I showed like years ago how to fix it yourself by disabling the send back to the return. I remember getting a project displaying the issue, fixing it and posting the solution.

I guess Ableton could automatically disable the send of a track to a return that is routed to the same track, but they probably assumed the people that would do stuff like that would be knowledgeable enough to see what to do about it, and if you can live with the extra sample of latency and possible phasing you can do interesting loops.

I'd rather Ableton did not babysit us. But they should be better at dispensing vital info that is actually not obvious for people not reading the fine-print if the manual or been in these forums for some time.

If other DAW apps don't have this issue, that's because they prevent loops from happening or they at least default to turning off the sends.

[2pauluzz2]

What do you mean fixed?

In Live 9 Ableton calculates delays from left to right, getting to the sends last, and if you then route the audio from the sends back into the project they will forever be late, causing a delay. I stopped using sends in Live for this very reason.

By fixed I mean: does this behaviour no longer occur in 10? We might not be talking about the same thing.

I showed like years ago how to fix it yourself by disabling the send back to the return. I remember getting a project displaying the issue, fixing it and posting the solution and then no-one bothered to post anything further.

Can you point me to that explanation? I've never been aware of the discussion/solution you speak of.

[Stromkraft]

In Live 9 Ableton calculates delays from left to right, getting to the sends last, and if you then route the audio from the sends back into the project they will forever be late, causing a delay. I stopped using sends in Live for this very reason.

This is not correct. All tracks are calculated per one sample. If they were not tracks couldn't be nullified using the Live mixer. When you potentially set up sending back a signal to where it came from Live needs at least one more sample round of processing. This causes phasing in some material.

It's somewhat odd you stopped using sends for this reason as typically you do not need to route a return to an audio track. And if you do, all you need to do is disabling a send.

By fixed I mean: does this behaviour no longer occur in 10? We might not be talking about the same thing.

We are talking about the same symptom and setup as far as I can tell.

Can you point me to that explanation? I've never been aware of the discussion/solution you speak of.

What part of my advice to "disable the send going back to the return routed to the track" is unclear here? That's all you need to do. But I'll see if I can dig it up.

[Stromkraft]

Can you point me to that explanation? I've never been aware of the discussion/solution you speak of.

Please read these two posts and maybe the whole discussion about a problem, which solution should also solve what you're asking about.
>>>> Setting up a drum bus??

OK, so I've found also a simple technique I overlooked when reading up about this situation: simply deactivate the sends on the drum buss by right-clicking on them and selecting that option. This will improve audio somewhat. In my quick newer experiment involving Utility on the return as the only effect it did not make polarity cancellation occur, but sound was certainly improved with or without Utility active (reversing the polarity or phase of the signal). EDIT** As you see in the next post this wasn't really true. I got these initial results because of different levels, so this first test was flawed ***

Again it was info from Nico, this time from 2011 (My bolded sections):

"There will be latency (*) on a return track if you route the return track's audio back into another audio track. The following problems will occur:

1. Such a routing can cause an infinite feedback loop, because you could (in theory) use the audio track's SEND controls to feed the signal back into the return track and so forth. In this case, Live cannot compensate latencies anymore, so the latency compensation will be deactivated for all tracks which are involved in this feedback network, regardless if the SEND control is indeed used by you or not. A possible solution is to deactivate the SEND controls on the audio track (right-click on the SEND control and choose "deactivate").

Note that the same problem may occur when you activate the SEND controls on the other return tracks, e.g. activate SEND B on RETURN A and SEND A on RETURN B.

2. There will always be at least 1 sample delay when routing a return track back into an ordinary audio track. This has to do with the computation cycle of the audio engine. Here's a (simplified) explanation: Live basically "scans" each of your tracks from left to right and "collects" each track's audio data, sample by sample. At the very end of each cycle, the master track's audio (or anything that is routed to your soundcard's output) will be sent to the DAC. Then it starts over, calculates the next sample and so forth.
Now, if you route audio from a return track back in to an ordinary audio track, this signal will always be delayed, because it falls into the next computation cycle. Thus you will never be able to reach full phase cancellation in such a routing scenario. There is no workaround for this. Note that this really only affects return tracks which are routed back into audio tracks. Routing an audio track into another audio track is no problem, regardless into which direction.

…

The next post was when I found the actual solution:

Follow-up:

I can report that I totally removed time-based phase issues with removing sends on any buss that the audio is routed trough in this simple experiment:

Source MIDI instrument 1 SynthDrum routes to 2 Drum buss which routes to Master. All of these faders are on xdB position.

SynthDrum sends 100% to the return track Utility Reverse (containing Utility switching polarity of the signal) with xdB fader position, which send its signal either to Master or Drum Buss, doesn't matter which. There is total cancellation provided the buss sends are disabled also when routing the return to the drum buss. If sending return to master it doesn't matter.

I'd say problem is solved! What do you think?

In relation to this Brian, I think, shared a project displaying these issues that I fixed with the technique. I also use this kind setup, like you, somewhat often myself in different situations, but not in every song.

[Stromkraft]

Looking over the solution, while it's certainly working beyond doubt in my experience, I realize it would be quite useful if Live defaulted to disabling the send back to the return whenever a user routes a return to a track.

If later the user attempts to enable this send knob, that ought to be allowed, but you should also get a warning.

[2pauluzz2]

Awesome, thanks. "Disable the send" didn't click with me initially- this is helpful.

[Stromkraft]

In Live 9 Ableton calculates delays from left to right, getting to the sends last,

This is not correct. All tracks are calculated per one sample. If they were not tracks couldn't be nullified using the Live mixer.

Reading over my Nico quote it seems that's more in line with what 2pauluzz2 says here. So

"There will always be at least 1 sample delay when routing a return track back into an ordinary audio track. This has to do with the computation cycle of the audio engine."

--Nico 2011.

That said I get total cancellation when I did my tests in 2015 and I have never experienced any phasing issues routing a return to anywhere as long as there
is no signal going back to said return. Total cancellation when comparing two signals with one being polarity reversed means both signals are
phase locked and also exactly the same. Of course there could be some issues in the unhearable range or so low I didn't notice it, but that'd be mostly academic knowledge.

Update:

I've redone the test from 2015 now in both Live 9.7.5 as well as 10beta, expanding on it a bit. AFAICT, the info from Nico is dated and Live 9 fixed what he was talking about. My correction of 2pauluzz2 therefore stands.

Here's the test:

Two separate systems before the master buss:
3 audio tracks are routed to a buss. These audio tracks are also sending fully to a return with twentysomething EQ8s in oversampled mode to get some latency to compensate.
The return is also routed to the buss.

The other system, track 5 to 8, is exactly the same, with the exception its return has more EQ8s. All in oversampled mode.
The buss II track (not the return) also has Utility with preset "Phase Imvert" (Polarity Inversion).

Because both busses contain the same signal these cancel out completely. This means the returns can't be even a sample late, as that would leave some signal.
As you can see in the screen dump there is no signal in the Spectrum analyser in the master track (due to cancellation).

You can check this is true by (I can share via Splice if needed):

• Enable the send in one of the busses, to the return routed to it. This prevent proper delay compensation.
• Delay one of the returns 1 sample with Track Delay. This illustrates how it would sound if the return was late.
• If you want to listen to the effects without nullification simply solo one of the busses or play only one of the scenes.

The Test set (Live 10):

[strangedaysuk]

Ok I dont post much here but on seeing this post I just needed to chime in.

Now I'm not some big pro mixer - but I have had the good opportunity to work with some top producers/mixers (doing chart material) and learnt the craft "hands on". My mixing skills improved a lot and so did my ability to be a mix engineer. By this I mean I can sit at most mixing platforms and hardware desks and once I work out the technicalities I can focus on the art and come up with reasonable sounding mixes from what I've been told. So I've used Pro Tools, Logic and Ableton as well as many others with PT being my strongest. I bought into Ableton as it suits my writing desires.

Is there a sound difference - perhaps but I didn't have any issues mixing in Ableton. Yes its very different but I ended up with results similar to Logic - maybe better as I was inspired in a different way. But to suggest that and DAW today has show stopping audio summing is crazy. In fact I've noticed the workflow brings different elements.

Having plugins like the Glue compressor built in as well suits me I always used an SSL style comp but having glue is so quick and easy and it works. I use third party plugs mainly but the basics such as the EQ and simpler are great to have in there.

So am I a "fanboy". I love them all. I use PT for live bands, Logic for songwriting and mixing and now Ableton. Not once has the audio engine in any of them concerned me. Plenty of issues exist in 9 (some sorted in 10 and yes I am annoyed at the upgrade price cause all I really wanted was better dPi scaling).

So if you have to mix in another DAW sorry about that! I find mixing in Ableton to be a breeze.

[pencilrocket]

I have occasionally experienced that audio clip playback sound different when using scrub area in Arrangement View. But I can't verify the rendered song is affected by this. It may happen only when realtime playback.