Live 9 NEEDs to fix SRC algorithms

Share your wishes for the future of Ableton Live
Post Reply
djcl.ear
Posts: 22
Joined: Fri Jul 02, 2010 5:12 pm

Live 9 NEEDs to fix SRC algorithms

Post by djcl.ear » Sat Sep 07, 2013 11:14 am

Yes, Live 9 does deviate the sound output of SRCs (Sample Rate Conversions) in unacceptable levels as it is.
Check http://src.infinitewave.ca/ precise measures, comparing Live 9 side by side with most other DAWs and most importantly, comparing its results with the expected behavior of an ideal filter, at least in terms of phase.
Live 9 measured error is too gross as it stands now.


djcl.ear
Posts: 22
Joined: Fri Jul 02, 2010 5:12 pm

Re: Live 9 NEEDs to fix SRC algorithms

Post by djcl.ear » Sat Sep 07, 2013 6:11 pm

No I am far from satisfied, and I suspect most needing to do any resampling work on LIve won't be either, if they find out that Live 9.03 shows -according to very precise measures- a deviation of up to 180° in phase. Deviation that does sweep exponentially depending on frequency, and starting by 4 kHz and up to 22 kHz.

Please go back to http://src.infinitewave.ca/ and check the graph for Live 9.03, test results for Phase, and compare the deviated green line shown, against the while line that depicts the ideal result from a perfect filter.

You might also take the time to see that Protools, Cubase and even Audition, Audacity 2.0.3 or FL Studio 11 get both, green and white lines, perfectly aligned... as opposed to Live 9.03 that shows the mentioned gross disparity.
Ableton needs to correct this situation soon.


PD: the test credentials are described in the site "Credits" section, and have no contest in this field.

Angstrom
Posts: 14722
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Re: Live 9 NEEDs to fix SRC algorithms

Post by Angstrom » Sat Sep 07, 2013 6:19 pm

So, you are asking them to fix the SRC, and they have fixed the SRC in the upcoming version 9.1 by integrating an SRC which produces a trace which now looks like this

Image

And you are still unhappy about this new SRC, which is what you asked for them to fix so that it looks like that?

Am I understanding you correctly? You are unhappy that they fixed it?

Go to src.infinitewave and take a look at the phase traces for SoX , because that's what they are moving over to.

Tarekith
Posts: 17634
Joined: Fri Jan 07, 2005 11:46 pm
Location: Ableton Forum Administrator
Contact:

Re: Live 9 NEEDs to fix SRC algorithms

Post by Tarekith » Sat Sep 07, 2013 8:17 pm

Wow, that line is not straight at all!

(Has nothing to do with SRC, just pointing that out)

djcl.ear
Posts: 22
Joined: Fri Jul 02, 2010 5:12 pm

Re: Live 9 NEEDs to fix SRC algorithms

Post by djcl.ear » Mon Sep 09, 2013 4:16 am

Angstrom wrote:So, you are asking them to fix the SRC, and they have fixed the SRC in the upcoming version 9.1 by integrating an SRC
Ok, now I understand... You are talking about upcoming version 9.1

I didn't imagine someone could be talking here about the unreleased version.
I plainly understood you were referring to version 9.01, and this when the mentioned tests were applied and published regarding version 9.03.

//So, yes: GREAT to see that the issue was -rapidly indeed- solved by Ableton, even without us users asking here.
And thank you Angstrom for the disclosure or the citation, if the information about the upcoming improved SRCs in Live 9.1, was published somewhere.

djcl.ear
Posts: 22
Joined: Fri Jul 02, 2010 5:12 pm

Re: Live 9 NEEDs to fix SRC algorithms

Post by djcl.ear » Mon Sep 09, 2013 4:49 am

Furthermore,
The Sox Algoritms seem perfect in terms of phase, very much the same happens with most other DAWs in the infinitewave tests, however regarding the precision of 'transition' is vastly lower in SoX than comparing with Izotope's results...
Angstrom wrote:they have fixed the SRC in the upcoming version 9.1 by integrating an SRC which produces a trace which now looks like thisImage
Angstrom, the image you posted here is not about Phase, it is about measuring Sweep, which incidentally in SoX graphs shows more reflections (not good) than comparing (again) with Izotope...

The Infinitewave tests are quite novel in that it is the first time such broad meticulous measuring effort is gathered regarding DAW's sound quality.
//in sum, Ableton like other developers needs to take note and hopefully take action to keep improving in this respect, because despite differences might seem minute in the numbers, they do reflect algorithm's quality DO have variations... and more importantly; doing what a DAW does, involves the application of many DSP processes in which these algorithms are multiple times applied and thus... their flaws might additively come to be perceivable, precisely because their flaws occur and show always in the same frequencies.

[mja]
Posts: 17
Joined: Mon Aug 15, 2011 12:39 pm

Re: Live 9 NEEDs to fix SRC algorithms

Post by [mja] » Mon Sep 09, 2013 8:57 am

The phase will look like this in 9.1
Image

Tarekith
Posts: 17634
Joined: Fri Jan 07, 2005 11:46 pm
Location: Ableton Forum Administrator
Contact:

Re: Live 9 NEEDs to fix SRC algorithms

Post by Tarekith » Mon Sep 09, 2013 2:18 pm

Does it have to be green, can we get a blue option for slower songs and ballads?

Angstrom
Posts: 14722
Joined: Mon Oct 04, 2004 2:22 pm
Contact:

Re: Live 9 NEEDs to fix SRC algorithms

Post by Angstrom » Mon Sep 09, 2013 3:00 pm

Tarekith wrote:Does it have to be green, can we get a blue option for slower songs and ballads?
Got dem SRC jokes

(????)? ¯\_(?)_/¯

Post Reply