Is Digital Performer Audio and/or Summing Broken?

[hotshot]

Originally Posted by dannygold

There's a lot of negativity applied to the concept of evolution too. Doesn't make it wrong.

The 'theory' doesn't make it right either. The null-thang is flawed:

I challenge you null and 'all DAW sound the same' guys & gals to show that when summing using the same plug-ins (both FX and VIz) that different DAW programs will null out. Here's a simple test: do a BFD with a Ren Comp and print the exact same thing in DP and Cubase. Make all settings, levels, etc. the same. My theory: They will not null.

The null test, IMHO, are flawed because they don't take plugins into account. Who doesn't mix with plug-ins if they are not using external summing and/or outboard gear.

Danny: can you tell us if DP (4.x) sums using 24 bit or 32 bit? Also: has the summing and/or plug-in integration algos changed between 4 and 5?

Good luck,

hotshot

[santibanks]

Originally Posted by hotshot

Danny: can you tell us if DP (4.x) sums using 24 bit or 32 bit? Also: has the summing and/or plug-in integration algos changed between 4 and 5?

Good luck,

hotshot

Audiodesk 2 says: record and edit @ 24 bit, mix @ 32 bit. So I think the engine is 32 bit.

[grimepoch]

Exactly how is it flawed? Your argument makes absolutely no sense. Obviously if you add plugins and changed and write out two different audio projects and you bring them together and do the null test, if they do not null each other out, like you've said, then THERE IS A DIFFERENCE. That's the whole point!

Let's be clear, the question here was about the SUMMING part of the DAW. Not the handling of gain, panning, plugins or anything else. There are other parts of the DAWs that are different, and no one is arguing that. I spent a long time demonstrating on graphs about the pan differences between Logic and DP and, as I said, creating a plugin for those that wanted to create their own pan curves.

My test CLEARLY and CONCISELY shows that between Logic and DP, the summing engine works IDENTICAL. They both use 32-bit float and they both have used 32-bit float for a long time. Just like almost all other total in-the-box DAWs. As long as we've used 32-bit computers we've had 32-floats and there has never been a reason NOT to use 32-bit so you get the power of the FPU.

Even further, I surmise that if the summation of signals on the mix bus is less than 24-bit, as most do since they are never in the red, that even with an HD system you are going to see the same results. We are within the resolution. Where I am VERY curious is when the mix bus peaks, I want to see if with the HD if the values are different around that point.

Note however, this is NOT an argument that 32-bit is more accurate than 48 bit INT because we know it's not, especially things for low pass filters. There are many white papers on this topic and it's pretty clear that negatives of floats are a lot higher than ints at these sizes. The ints do not deal with all the same round-off issues. Again, I am only curious in the summing.

And as a final thought, I have NO doubt that since the summing is the same between DP and Logic that effects will be the same since they both use the AU protocol and they both use the same internal data structure.

[grimepoch]

1) Panning hard right and hard left DOES NOT ensure the same operation in all DAWs. Also, MONO sources are not handled the same either because (a) Center panned mono signals are affected by the Pan Law. Since the mono signal is going to both left and right, the amount of attenuation of that signal is TOTALLY dependent on which pan law settings are used.

For instance, some use -3.5 for center, other -3.0. On mixing consoles I have seen -6dB for mono sources at center gain. DP and Logic use different pan laws but THIS DOES NOT AFFECT DYNAMICS. It just means if you are using a mono source in your test, or are panned all the way left or right, you need to know that your levels are. This information I believe can be obtained from the documentation, but also, is easily tested using a SIN source sample at a specific DB setting and just adjusting the pan values and looking at the intensity on the specific left, right, or both on a mono source.

Stereo panning is not really panning but balance of the left and right channel. When you turn the pan, the opposite channel to where you turned gets reduced. This reduction in DP is linear, and in Logic is logarithmic. The same setting in Logic will APPEAR wider because less of the signal is on the other channel when compared to the same setting in DP.

2) I am working on a test with FX plugins in both Logic and DP. What I will tell you about VIs is that there are two separate problems that I am dealing with at the moment.

a) Not all VI's reproduce the same, they have time dependent characteristics that change over time for the same frozen location of a song. I recorded a passage of Microtonic 4 times in each DAW and on each pass the results had differences in them ( at the sample level, the waveform was NOT the same) and this is in the SAME DAW. I used another VI, MX4, and the samples were EXACTLY the same on all four passes. I suspect with BFD that since drums tend to sound better with a bit of human or random feel to parts of the sonic quality that if you bounced the same VI of BFD in Logic 4 times, the output would NEVER null out, and it shouldn't. The design of the plugin prevents the output from being identical. In my case, I am going to use WaveArts reverb and Izotope Ozone3 on a few tracks in each and see what happens.

b) I have discovered a problem in both DP and Logic that VI instruments are not getting fired at the right times. In both programs they have different problems. Some other people have indicated they see this as well, so I am trying to track it down. If I bounce a MIDI piece in both DP or Logic and look at that audio in relation to the MIDI driving the synth, I find that the recorded audio is late AND not consistent. In the case of Logic, the first note is WAY early and also causes a pop because it starts at the max amplitude of the first note. Not good.

In any case, more to ponder. FX tests coming soon. And really, does NO ONE have access to an HD system to try out the test files?

[dannygold]

Danny: can you tell us if DP (4.x) sums using 24 bit or 32 bit? Also: has the summing and/or plug-in integration algos changed between 4 and 5?


DANNY SAYS:

Definitely 32 bit float in the internal processing engine. I doubt the summing or plugins changed (at least sonically) between 4 and 5. But I didn't work there when 5 came out so I couldn't say.

[andremattos]

Originally Posted by grimepoch

Strange, I am on 10.5.2 as well.

In any case, as you requested (I forgot about the compress in Finder)

http://www.asylumstudioproductions.c.../audioTest.zip

Rick

Logic and DP, the summing ? IDENTICAL!
i cant hear any diference!

[grimepoch]

As a note here, while most of this discussion is happening over at unicornation at the moment, I wanted to tell you all I have been experimenting with FX plugins HEAVILY at the moment.

For time dependent value information (filters, compression) the results out of DP and Logic are different, dynamic wise on compressors even, and I am trying to figure out why. Things like gain in a plugin work as expected (null to zero).

I am in the process of looking for or while have to write a plugin to examine what the host is asking of the plugin. This way, I can see if this is a time scale issue, time slice issue, or what.

Hold on to your pants! Interesting stuff.

[andremattos]

Great!
Please , Keep us posted!
thanks

[newrigel]

Originally Posted by Kazrog

I was a DP die-hard since 1997 until a few months ago when I did some summing tests between it, Pro Tools 7.3, and Logic 8. I've since sold my MOTU gear, got an M-Audio Project Mix I/O and Pro Tools M-Powered, haven't looked back. I can't wait to get my band's next record out, even our rough songwriting demos are sounding better than our last album!

DP's summing is really bad, and my mixes always sounded crowded and muddy, and it took hours of fighting this and weird mixing tweaks to try to overcome. Both Pro Tools and Logic have infinitely better summing than DP based on some of my tests where I was pulling up DRY tracks with NO plugins, just some volume and panning adjustments matched between applications with the exact same audio files.

MOTU needs to get it together, they have been really evasive, defensive, and even mean to their loyal customers who have asked them about the summing issues. Everyone I know who is a professional agrees with me about DP's summing being awful, and MOTU is in denial. Their application also hogs processor resources more than any other software, for no apparent reason.

Oh yeah... DP 6 is out so we'll soon see!
I get great mixes out of DP and it's all how you go about it. No DAW out there can have the busses slammed so you have to watch your gain stages and know what your doing to get the imagery and definition out of the mix...

[andremattos]

I really like DP5!!
As i told , i was PT user for 8 years , and now i'm recording / mixing in DP5
Now my mixes is getting better ..
I love DP!

[jeronimo]

Originally Posted by grimepoch

As a note here, while most of this discussion is happening over at unicornation at the moment, I wanted to tell you all I have been experimenting with FX plugins HEAVILY at the moment.

For time dependent value information (filters, compression) the results out of DP and Logic are different, dynamic wise on compressors even, and I am trying to figure out why. Things like gain in a plugin work as expected (null to zero).

I am in the process of looking for or while have to write a plugin to examine what the host is asking of the plugin. This way, I can see if this is a time scale issue, time slice issue, or what.

Hold on to your pants! Interesting stuff.

It's clear to me that character plugins like dynamics, etc, will sound different.
Try some AU plugs like the sonalksis compressor, use the same settings and see if they'll null.
I really have no idea... but I'm curious to know the result.

[grimepoch]

I tried a variety of plugs and I do not think they will null. All I can think is the timing information from the host must be either (a) in different intervals or (b) at different resolutions.

Regardless, I am talking now with CoreAudio people (well, waiting on a response really) to see if I can get some light shed on these issues.

[jeronimo]

Originally Posted by grimepoch

I tried a variety of plugs and I do not think they will null. All I can think is the timing information from the host must be either (a) in different intervals or (b) at different resolutions.

Regardless, I am talking now with CoreAudio people (well, waiting on a response really) to see if I can get some light shed on these issues.

or maybe the plugins doesn't act in a "linear" predictable way... which I don't believe is the problem.

[grimepoch]

Believe me, I am just as clueless about this at the moment.

I do know that on certain hosts, many times a plugin will get data stuck in it. Like, when you rewind, start a song and the reverb starts ringing out what was sent to it before at some other part of the song (which is an absolute pain when you are bouncing out, always play some silence first).

Something like a compressor should be pretty straightforward. All I could imagine is that it might depend on the chunk of audio the plugin is asked to process. I know that Logic uses a lot of dynamic buffer sizing and DP uses a lot of pre-rendering for speeding up performance. Although, changing buffer sizes in each on deterministic plugins had the exact same results in the same DAW. Given that if I just use a gain slider in a plugin the results still null out, it's not a matter of some strange hand-off of the floats. I am thinking that the size of data passed in (like process these 128 samples) must be different OR that something is up with the unit of time.

I am more concerned with the fact MIDI notes do not consistently fire VI notes at predictable times in either DAW. (Of course, truth be told, external MIDI has this problem depending on how much data you are sending out your MIDI interface as well, and this is a known and unavoidable thing).

[pipelineaudio]

zero out the plugs and run a transient thru, see if it isnt a timing issue (which I wouldnt be surprised about at all)

[grimepoch]

The sad part is that it is not consistent.

I created a sin @ 1khz at 0dB. I made this segement one bar long, then a bar of silence inbetween.

I then took this and fed it through the Apple Dynamics AU with the Hard and Fast setting (default).

When I nulled both files, not only was there a difference, it was NOT consistent on each of the bits of audio, they were different and the resultant different waveform had different shapes! The first one was sinusodial, the third one was more sawtooth shaped.

Something time wise is not the same between the two. Extrapolating the cause is turning out to be rather difficult.

[jeronimo]

Originally Posted by grimepoch

The sad part is that it is not consistent.

I created a sin @ 1khz at 0dB. I made this segement one bar long, then a bar of silence inbetween.

I then took this and fed it through the Apple Dynamics AU with the Hard and Fast setting (default).

When I nulled both files, not only was there a difference, it was NOT consistent on each of the bits of audio, they were different and the resultant different waveform had different shapes! The first one was sinusodial, the third one was more sawtooth shaped.

Something time wise is not the same between the two. Extrapolating the cause is turning out to be rather difficult.

Oh boy I just love this!

This brings me back to what I always tried to say. Suming probably is the same on all 32 bit float DAWs, but this doesn´t mean nothing... cause EVERYTHING else is different.

[grimepoch]

I got an answer and am now in the process of building an AU for testing. It probably wont be till Monday before I get some good results. Mainly, I need to code up some other code to parse the data I generate.

Hold on to your shorts!

[hotshot]

Originally Posted by jeronimo

Oh boy I just love this!

This brings me back to what I always tried to say. Suming probably is the same on all 32 bit float DAWs, but this doesn´t mean nothing... cause EVERYTHING else is different.

EXACTLY! That's why some may sound better then others on a VERY noticeable level.

Good luck,

hotshot

[Oroz]

Originally Posted by drBill

A unity gain mix in PTHD sounds significantly better thru identical converters than it does in DP.

Inspired by this thread I did a mix with no plug-ins, just levels and panning, in Pro Tools HD 7.4 and then did the same mix with the same level and pan settings (all hard panned or dead center for consistency's sake) in DP 5.11.

Both mixes (with 30 tracks BTW) sounded the same to me so I went further with the test. I put both mixes side by side and inverted the polarity on one of them. Guess what? They perfectly cancelled out! I was expecting some degree of cancellation but not 100% do you know what I mean? This proves that at least DP and Pro Tools sound exactly the same.

[grimepoch]

Thanks very much for doing that test, I've been wondering. I am still analyzing other issues with some customer AUs I am writing, but this test was VERY important to me.

[kubimusik]

Originally Posted by Oroz

Inspired by this thread I did a mix with no plug-ins, just levels and panning, in Pro Tools HD 7.4 and then did the same mix with the same level and pan settings (all hard panned or dead center for consistency's sake) in DP 5.1.1.

Both mixes (with 30 tracks BTW) sounded the same to me so I went further with the test. I put both mixes side by side and inverted the polarity on one of them. Guess what? They perfectly cancelled out! I was expecting some degree of cancellation but not 100% do you know what I mean? This proves that at least DP and Pro Tools sound exactly the same.

I'm assuming this was DP running under DAE, correct?

[Tony Shepperd]

Originally Posted by Oroz

Inspired by this thread I did a mix with no plug-ins, just levels and panning, in Pro Tools HD 7.4 and then did the same mix with the same level and pan settings (all hard panned or dead center for consistency's sake) in DP 5.1.1.

Both mixes (with 30 tracks BTW) sounded the same to me so I went further with the test. I put both mixes side by side and inverted the polarity on one of them. Guess what? They perfectly cancelled out! I was expecting some degree of cancellation but not 100% do you know what I mean? This proves that at least DP and Pro Tools sound exactly the same.

No offense, but this is like testing your cars performance while idling.

[delcosmos]

Originally Posted by Tony Shepperd

No offense, but this is like testing your cars performance while idling.

Why?

He is proving that both systems sound the same.

I think this is a great test (Thank's Oroz), I remember somebody saying that same files sound different on PT than DP.

Best regards.

delcosmos.

[Tony Shepperd]

Originally Posted by delcosmos

Why?

He is proving that both systems sound the same.

I think this is a great test (Thank's Oroz), I remember somebody saying that same files sound different on PT than DP.

Best regards.

delcosmos.

If you have two cars.
One engine was designed for high performance and the other wasn't.
But you test them both while the engines are idling or not running at high performance.
Then it's not an extensive test.

If you mix in the box, what starts to sonically clog up your system is processing.

IMO, the more processing and plugins your throw into the system, is where it reveals it's flaws.
How well it does under sonic pressure of a lot of tracks with a lot of processing.

Just like an engine does when it's performing at it's peak.
That's one true mark of a really great performing car.


And more importantly (to me), that kind of real world test, is where having accel cards in PT HD, gives you that extra boost of sonic power.

Of course, YMMV.

[delcosmos]

Originally Posted by Tony Shepperd

The reason is, if you mix in the box, what starts to sonically clog up your system is processing.
IMO, the more processing and plugins your throw into the system, is where it reveals it's flaws.
How well it does under sonic pressure of a lot of tracks with a lot of processing.

Just like an engine does when it's performing at it's peak.
That's one true mark of a really great performing car.


And more importantly (to me), that kind of real world test, is where having accel cards in PT HD, gives you that extra boost of sonic power.

Of course, YMMV.

Ok, but I feel this is a valid test because some people stated here that same files without any processing at the exact same levels sound better on a PT session than on a DP session. This test proves this is wrong.

I mix mainly with outboard gear, I use my DAW just as a recorder/playback machine.

I know and accept that PT is the standard and yes, it is really powerful, but I am talking sound quality here (no processing).

Best regards.

delcosmos.

[pigcat]

Correct me if I'm wrong but...

Aren't summing engine is NOT the same as sound engine?

You need whole bunch of tracks to "sum" together, but you're testing sound engine by just playing back, even just 1 track.

I find, if you take the same set of audio and do the test, yes, there aren't going to have a difference more than 5% - or subjectively each have their own pros and cons.

However, take a bunch of virtual instruments and do the test again. You don't even need to sum the tracks, you'll be surprised, even by just hitting the first note...

[delcosmos]

The funny thing is reading a lot of people here claiming that DP sounds TERRIBLE!

They are completely sure that just running the same files on DP and PT results in a dramatic difference (no plug ins or processing, just levels and hard pans ).

This is why Oroz's test is so revealing.

BTW, I own both systems, a PT HD3 with an Aurora 16 and control 24, but I work EVERY DAY with a DP 5.12, Apogee AD16X(x2) and DA16X(x2) and a SSL Duality with a big selection of outboard gear and I just can't produce on PT, I'm so used to DP. Our Pro tools mixes sound great, but NEVER better than our DP mixes.

Best regards.

Delcosmos.

P.S. what a great thread!!

[pigcat]

Originally Posted by delcosmos

Our Pro tools mixes sound great, but NEVER better than our DP mixes.

Were you trying to compare ITB mixes with OTB mixes?

[delcosmos]

Originally Posted by pigcat

Were you trying to compare ITB mixes with OTB mixes?

No!

Just OTB or ITB mixes with two different DAW's (PT and DP).

I believe that a good engineer can mix ITB or OTB if he knows his tools, there are ITB mixes that kill OTB mixes and OTB that kill ITB mixes, in the end it's the driver not the car.

All my discusion here is about the revealing test by Mr. Oroz where the same Audio files sound exactly the same on DP and PT mixes. Honestly I'm enjoying this