Live audio quality?

[abort]

it's 64 bit summing, not across the board 64 bit processing.

the change was something like a ~-80dB lower noise floor or somesuch nonsense.

you can use typical dB calculations with floating point numbers because the numbers can represent numbers between 0 and 1 or between 100 and 3,000... whatever Live needs at the time.

I think the dB calculation is your sound card setting, that's where Live converts 32 bit floating to a fixed point value, at all points before that up to your sound card it's floating point.

Oh shit! and what 'setting' would that ~ to in a typical sound card?

I'm trying to understand really!!

[abort]

oh f***k I guess ../MORE HOME HOMEWORK for me.

[Tone Deft]

it's 64 bit summing, not across the board 64 bit processing.

the change was something like a ~-80dB performance gain or somesuch nonsense.

you can't use typical dB calculations with floating point numbers because the numbers can represent numbers between 0 and 1 or between 100 and 3,000... whatever Live needs at the time.

I think the dB calculation is your sound card setting, that's where Live converts 32 bit floating to a fixed point value, at all points before that up to your sound card it's floating point.

Oh shit! and what 'setting' would that ~ to in a typical sound card?

I'm trying to understand really!!

bad post by me, fixed above.

I can't tell if you're being sarcastic or what, my sarcast-o-meter is worn out.

basically it doesn't matter, there are more important things to learn about, like how to properly use mastering tools.

[hacktheplanet]

Must be a phase issue. The 65ms at 5.74 Mdhz can only be explained with a seventh order mercury shift. And this is CPU specific, and no software can change this. I bet you'll find a similar ondulation on a PC too.

However, I have serious doubt if this would be audible. If you simply fire a clip using the mouse, you'll create a Columb field of around three Nanowebster, unbiased (!!!). Via MIDI the results will be slightly better, around one or two Nanowebster, which is still higher then what you'll get using a medium sized Cembalo.

Thanx
R.

[abort]

No I'm for real. Just wanted to figure all this out.

Vary few times I would be sarcastic.

[Tone Deft]

kewl, I like this stuff, talking about it helps me reinforce what I know and find out what I really don't.

try reading this wiki entry, then ask questions.
http://en.wikipedia.org/wiki/Dynamic_range

re: summing... it's the parts of Live where two or more signals are mixed together, they're done with 64 bit precision rather than 32 bits, so the rounding errors are lower.

you can also try searching for posts by "Robert Henke" on the word "summing" the guy's been very generous about laying it all out, or rather Ableton has been very cool about letting him post details on Live.

[abort]

Good!! I'll check this out. always something to learn here!

peace!

[Crash]

it's 64 bit summing, not across the board 64 bit processing.

re: summing... it's the parts of Live where two or more signals are mixed together, they're done with 64 bit precision rather than 32 bits, so the rounding errors are lower.

Could you explain please, how summing two 32-bit numbers gains additional 64 bit precision (less rounding errors)?

Simplified example of how I understand things:

32-bit precision output: 1.23
64-bit precision output: 1.23456789

32-bit precision outputs to 64-bit sum:

1.23 > 1.23000000

64-bit sum from 32-bit outputs:

1.23000000 + 1.23000000 = 2.46000000 = 2.46

Only (constructed) usage scenario of a more precise 64-bit sum while only 32-bit processing is used:

32-bit: 1.23 - 0.22 Fader Gain = 1.01 - 0.01 Fader Gain = 1.00

64-bit: 1.23000000 - 0.22441234 Fader Gain = 1.00558766 - 0.00050000 Fader Gain = 1.00508766 ~ 1.01

So the my conclusion is: As long as processing takes place in 32-bit the 64-bit summing is nearly useless. It's a marketing gag that conceils that Ableton didn't deliver the real-deal. Does it matter? No, because there are other things happening in Live, like phase shifting, that are far more detrimental to Audio Quality!

Sonar and Reaper come with 64-bit processing all way through, which is a different story. Since in processing there is a whole lot more calculations being done than in simple summing the higher precision promises at least some benefits. The more processing/number crunching happens the higher the benefits. And only once you sum tracks that use 64-bit processing themselves there will be an advantage in using a 64-bit sum.

[sweetjesus]

my vagina in seems deeper and wider

[tamtam]

So the my conclusion is:

It'll take Timur 32766 more nick changes to finally understand how it works

Maybe then he can come up with a less braindead example of rounding error, where more signals than 2 are involved or one where the error is actually accumulated, like common filter designs.

[Crash]

Maybe then he can come up with a less braindead example of rounding error, where more signals than 2 are involved or one where the error is actually accumulated, like common filter designs.

That's the point, tamtam: Summing is a braindead example of adding one 32-bit signal to the other! NO accumulated error, NO common filter design, NO benefits! Adding zeros at the end doesn't make it any more precise.

[Poster]

it's 64 bit summing, not across the board 64 bit processing.

re: summing... it's the parts of Live where two or more signals are mixed together, they're done with 64 bit precision rather than 32 bits, so the rounding errors are lower.

Could you explain please, how summing two 32-bit numbers gains additional 64 bit precision (less rounding errors)?

Simplified example of how I understand things:

32-bit precision output: 1.23
64-bit precision output: 1.23456789

32-bit precision outputs to 64-bit sum:

1.23 > 1.23000000

64-bit sum from 32-bit outputs:

1.23000000 + 1.23000000 = 2.46000000 = 2.46

Only (constructed) usage scenario of a more precise 64-bit sum while only 32-bit processing is used:

32-bit: 1.23 - 0.22 Fader Gain = 1.01 - 0.01 Fader Gain = 1.00

64-bit: 1.23000000 - 0.22441234 Fader Gain = 1.00558766 - 0.00050000 Fader Gain = 1.00508766 ~ 1.01

So the my conclusion is: As long as processing takes place in 32-bit the 64-bit summing is nearly useless. It's a marketing gag that conceils that Ableton didn't deliver the real-deal. Does it matter? No, because there are other things happening in Live, like phase shifting, that are far more detrimental to Audio Quality!

Sonar and Reaper come with 64-bit processing all way through, which is a different story. Since in processing there is a whole lot more calculations being done than in simple summing the higher precision promises at least some benefits. The more processing/number crunching happens the higher the benefits. And only once you sum tracks that use 64-bit processing themselves there will be an advantage in using a 64-bit sum.

Timur?

[Blomblom]

m u s i c M U S I C m u s i c 1 2 3 4 1 2 3 4 1 2 3 4
m u s t M U S T m u s t
b e B E b e
p r e c i s e P R E C I S E p r e c i s e
a n d A N D a n d
a c c u r r r r a t t t t e A C C U R R R A T T T E a c c u r r r r a t t t t e
1 p l u s 1 e q u a l s L O V E

[Crash]

Well, let's leave the irrelevant summing-engine behind and have a look at what really affects perceiveable audio-quality.

Everytime you launch a scene, press Play (spacebar) or run a "Loop in Arrangement" Midi plugins get shifted in time compared to Audio clips. That happens with both internal and external plugins and is independent of latency compensation. Some vary by a set ammount, some by varying amount (with new launch of the Scene, Playback, Loop). Under some conditions only the attack phase gets shifted while the sustain and release remain in sync.

The only time that doesn't happen is after the very first start of playback via the Play (spacebar) button after playback has been stopped via the Stop (spacebar) button.

Furthermore a similiar shift happens when playing Warp-Looped audioclips (at original tempo) vs. Envenlope Triggered audioclips. Just that in this case the shifts happen under different conditions (like after the first full loop).

As the more musically inclined among you might imagine that leads to unwanted phase shifting and modulation and can be detrimental to tight rythm/beat playback (like putting that Midi beat tightly in sync to that bass clip). But since most people here didn't ever notice it I guess it's a minor issue anyway and as such as irrelevant as the useless summing precision.

[leisuremuffin]

might as well sticky this thread too.


i just want to let everyone know that i have re-evaluated my position on this matter. I now believe that ableton is just a toy for dj's and will be doing all of my pro work in a dedicated pro environment made for pros and built to professional specs.

I have switched 100% to pantaloons superiour version mobius (pro edition) for all of my recording and mixing purposes and will now only use live for making light-brite style pictures with colored clips and epic trance dj sets.



maybe live 8 will have pro sound quality.


.lm.