Is the Total the Sum of the Parts..(Bouncing)

No, I haven't noticed a sound difference between midi and a bounced wav.  And they shouldn't be different - they should produce the same sound if you are bouncing to 24 bit files.  If you are bouncing to 16 bit there might be a slight degradation, but it will be the same for you CD.

@

Dear people,

Well, I just did another more controlled test, and this time I went back to an earlier project that had the MIDI and vocals recorded.

I wiped all previous mixes, and left the vocal and the MIDI tracks only.

I recorded an audio track "X" from the midi (by playing the MIDI and recording), and checked it to sound as close as possible in level to what I hear if I play and audition the MIDI. I can do that by mute unmute and solos. Audio track X sounds identical to the MIDI when played, in volume and quality.  Am using AKG 240 headphones, no speakers.

Then I mixed the vocal and the "X" onto a track "Q" by sonar-bounce function.

Then while PLAYING only the Vocal and the MIDI  track, I recorded another audio track "Z". No volume levels changed.

Now, since I had tracks Q and Z right next to each other, it thought why not compare by sight? So, I dragged their tracks so they have just about the same height.

Then I compared the waves. They look different. Track Z looks  proportionally and noticeable a little taller (more volume) BUT NOT ONLY THAT, the shapes are NOT identical in some parts, doing away with my initial thought that it'd be a volume difference. In shapes, it's much easier to notice and tell that they are not equal.

Of course, they do sound different, and Z, as far as I can tell has "better quality of sound", even if readjust the volumes, which were supposed to be normalized to begin with.

I think in this case, the ear can be easy to fool me, but not the sight.

I have checked carefully all track's inputs/outputs for consistency. The microphone is DOWN so there are no external sounds.

Any comments? Can some one be willing to test on his/her system this (just going to an existing project that has plenty of music/vocal content)? I am only using wave-in wave-out. I am using no buses, or special audio sends, since I don't use those anyway yet.

Or any comments or suggestions on why this is so?

Thanks.

what soundcard are you using?  does it have an EQ or other FX built into it that are being used without your direct interaction during the process of bouncing?

Hi all, again.

Mike-mccue, thanks for the comment which I consider correct, but I had already checked the waves at audacity too, by exporting the resultant waves, and they do differ.
--------------------------------

Beagle, thanks for your comment.
Due to it I went and checked and I did have a hardware FX for surround in my Audigy 2 ZS card, which I am not sure if it was affecting, because for a long time since learning about mixing and sonar, I have been using only Stero (not the 4 speakers)...And due to your comment, I re-did the test again and did still find differences still by inspecting the waves. I will do some more tests to see if that effect is active even if the system is set for stereo only, because that's supposed to be for when the speaker settings are set to 4 channels. I do consider your comment very pertinent, and I need to study it in case it is favorable, even, for quality or not, and so I know what I am doing and get control on these things.
    
---------------------------------------------

Thanks Kalle for the comment on synths..I do not unconsider it. I am using the Proteus VX and a Yamaha Soft Synth VSTi, and the Native B4...There are no loops nor things that I could attribute to viewable differences, though, but there could be.
****************
   
I DO HAVE this newer observation, after quite a few hours of testing and observing, after my previous posts, that are consistently reproduce-able:

Mainly, I can see differences in the areas where there is vocals.

In the vocal track, I DO HAVE  the audio effect plug-in "T-SLEDGE" which works only for playback, because I don't echo that track.
    
So, more and more, it does now seem to me, that the observable difference in shapes (in the areas of the vocals in the mix) seem to be due to the  way that the bouncing interacts with that audio VST effect.

So, maybe it is what that VST audio effect does IN REAL TIME (during playback), vs how it respond when called by Sonar to do a bounce that includes it.
    
I'll do some more tests with the same project later, without the VST on the vocal and report back.

I do thank the interest and help very much.



PS.: Slow bounce and fast-bounce of audio tracks made no difference, as expected.

Take all effects off your tracks then do your test. If the raw tracks without other processing are still coming out different, then it has to be your sound card processing.

If things come out OK then your VST is suspect but frankly, I have never heard of this happening.

An excellent thread for me!  (+1 to all)

For me it seems:

1) Bouncing is exact 'science':

2) Freezing tracks sucks in the Sonitus EQ and VST Fx's (Fx-bin)

3) Exporting tracks ads post fader reductions/boosts.

4) The summation of tracks is exact 'science' also.  If I doubled a 'clean' track (one without fx's, automations, MIDI coding, etc.) ... I'd 'expect' roughly twice the decibels, IIRC ... on the receiving buss.

5) But some synth samplers, like EWQL, Toontrack stuff, etc. vary samples and/or other parameters ... making it tricky for me to handle mixing sums/summations and such.

6) In Sonar 8.5.3, I have been oft unable to get consistent audio track outputs ... especially those containing Gain-clip envelopes and multi-splices ... unrelated to latency/buffers.  (IIRC, No other Sonarites have reported 'my' level problems with S 8.5.3)

In the past I've had suspicians with:
1) Soundcards (even my RME ff400)
2) DAW Motherboards (yes ... plus the fact I've messed with the registry a bit)

Don't know Sonar LE but is there a phase button that is accidently engaged? That would account for what you are seeing.

UPDATE-2
-------------

I did some more tests to try to find some clues or conclussions.

With a minimum of detail, it's like this:

1. With Sonar in MME mode, also the inversion results when playing a the above vocal wave (no FX) and recording while play in a separate track.

2. Play the wave track in Sonar, record it in Audacity while playing, ALSO results in a phase-inverted wave.

3. Furthermore, I removed the Audigy card rebooted and set system to play with the built-in chip which is, in my case a SoundMax HD chip. I verified the sound, played an mp3 at the Media player, etc.

4. In Sonar, using the SoundMax HD chip, MME driver mode, the inversion STILL IS THE SAME.

----
MORE INTERESTING THINGS:
==================
Stil using the built-in chip SoundMax:

5. So I exported the Vocal audio track out to wave.

6. Loaded it into audacity, inspected it there and concluded that it is correct (not inverted). So, Export is ok.

7. Then, this test is VERY INTERESTING: In Audacity, with such wave, I hit record, and I have audacity so it record a track underneath the existing one, while playing it..And, guess what? ..The RECORDED wave is INVERTED....In all these tests, I've learned to know the test wave, and where to check for a very clear difference. So I no longer need to do inversions to see that they are OK or inverted.

CONCLUSSIONS:

So,....something quite unexpected is happening in audio, as I find here:

1. Two recording softwares when playing an audio track and recording another in itself of the same thing, the resultant wave is phase-inverted.

2. The two recording softwares, do the same thing, even on two different hardware audio cards.

So which is wrong? Either BOTH softwares, or BOTH cards?

Can any one confirm in his system?  Now this ought to be easy to do, just need a wave form that has some observable ASYMMETRY.

It has nothing to do with vocal track, or with VST audio effects.

I hope someone can help on this and verify or deny this finding.

In the meantime, I'll check to see if I can find a test audio wave with some asymmetry on purpose, that would make this testing easier.

Thanks everyone.

NoKey, I'm not sure I see an answer to this question:

"2) I interpret what you are saying is that you are comparing a digital bounce to an actual analog "bounce". Is that correct?"


If this is the case I suspect you are simply observing that the analog summing is less linear than the digital summing. If that doesn't make sense search out the term "linear response" as it relates to transfer functions.

At least he's making 'recordings'.  I love his sci-method, though I hate clinical methods and procedures.

1) Now ... if he'd update his gear abit: card, hardware, Sonar, etc.

2) Get a spicy song going

Hi again, to all people that commented and have proper and understandable criticism:

I will simplify the whole thing again:

It turns out that in Sonar if one bounces one or more waves to antother track, the resultant wave will have say polarity "X".

However, if one plays the same waves simultaneously, while recording the result in another track wave "Y", the polarity of Y may be opposite to that of X.

Now, the fact that I did have some VST and even one inadvertent hardware effect, are found to have nothing to do with this.

But, why can this be relevant? It can be relevant to people like me or others who have a sound source which is not a VST and hence have to PLAY/record somethings and bounce others.

And of course this DOES NOT apply to recording exactly what one has (duplicating); but rather doing this to an existing track while adding some effect, or other things, or mixing.

The idea of recording the existing audio track, onto another identical wave in another track, while playing the source track, was used only to show the veracity that the resultant wave was indeed inverted in my system.

An example of where this can matter would be: Say you have a wave-a, and you want to add some hardware effect or software effect that forces one to play and record the result, rather than bounce it. And that somehow one wants to then bounce some of the original wave-a with the played-recorded result-b...Well, in that case obviously phase reversal could be important.

I am clear already to myself on what's happening, and specially after I found lastly the following:

It's a web site of a person who makes software to use sound-cards as an oscilloscope, so obviously he better be very familiar with waves, not only of the symmetrical type, but all kinds of waves, not just audio waves.

Well, he has this paraphrased comment: "There are some sound cards that INVERT the phase of the incoming signal BEFORE doing the digital processing".

Now, although I have no doubts on the fact that the two cards I tested INVERT the input signal, perhaps for others of us here, this might be of interest.

Otherwise, for most things it may not make much difference, or not matter at all. But if one does get some things by bouncing and others by play-record, then it might matter.

So, after all, the test is really very simple..A short vocal track or any wave with some assymetry can be used to play-record the wave, result in a second track (use no no vsts or effects), and just visually compare the resultant waves... A vocal wave with an intentional spike (like tap the mic a few times) can probably be a simple one to use.

Thanks Jeff, for trying to duplicate some of what I said at the start of the thread, where I was only starting to find this out.

But if you or others wish to report on this simple tests for phase inversion as said, it might be be interesting to know. No big project is needed for this test. Just open sonar and record a vocal with some spikes..Then play/record it into another track, and compare the two waves. If the spikes go up on one and down on the other, then the sound-card or sound interface reverses the polarity, otherwise it does not.

As said some cards do, and some don't. And why I don't know.

But I also read somewhere that some professional external DAC's do have a phase reversal switch. So maybe this is not necessarily a trivial finding or effort after all.

Thanks everyone!

Well, just in case, this clarification:

AT and Jeff, when I say Play-Record, all the sounds are internal, nothing is coming in via analog cables.

Jeff, I watched the video you linked. Very interesting. In the terminology of the Video, what I found is "polarity reversal" of the original wave vs the wave when recorded as played.

Also, the quality difference I referred to was due to a hardware effect being on, as Beagle suggested. I realize that phase shift is hard to hear. And also that a phase reversal by itself is not hearable, only important if one does some adding of waves for whatever reasons.

Thanks again.

NoKey,

That's a very thoughtful post.

FWIW, I Reamp guitar tracks frequently with my MOTU gear... I'll check tomorrow to see if the phase is inverted on the second pass. I don't think I've ever had an issue with it before and I usually use the direct track and the Reamped track in a blended mix.

I use the MOTU ASIO driver and always do a round trip check and manual latency compensation before doing overdubs.

best,
mike