Weird Signal Flow Artifact

[GrizzlyAK]

Hi, I'm working an old project in Audition CS6 with some complex routing that allows me to render either 2.0 or 5.1 output and I'm seeing a weird signal anomaly that I can't explain. I'm hoping someone here can shed some light. Basically, I built all tracks in the session to OUTPUT (or Send) to a 5.1 buss chain to the ultimate MASTER 5.1 track. To build the 2.0 Mix, I used Sends on each track to a 2.0 signal chain to ultimately, again, the 5.1 MASTER track. (when I bounce the 2.0 mix, I get a 5.1 file with information only in the L/R channels, which I then downcovert to a true Stereo 2-channel file, adding back in the 8.15 dB lost in the conversion from 5.1 to 2.0).

Here's my issue: I have a mono clip on a mono track that is OUTPUT to a 5.1 buss chain (which is muted). I have a Send on that same mono track to a stereo track that is ultimately routed to the MASTER 5.1 track. I have that Send panned 100% Left. When I play the clip, the meters from the track, all the busses, and the 5.1 MASTER show information ONLY in the Left channel (even with meters set to 120 dB range). That's what I would expect.

Now, here's the interesting part. When I mixdown the session where ONLY that one track is Soloed, I get a 5.1 file that has information in BOTH the L and R channels. The Left channel is very Predominate, as you would expect, but there is some kind of visible (and audible) "echo" on the R channel at a much lower volume. Since the Meters do not show this on playback, and I can't hear it on playback, WHY is the mixdown putting signal in the Right channel? See attached image.

Note that I do have clip effects on the audio (Au parametric EQ and a Hard Limiter), but disabling the entire Rack does not eliminate the echo (i.e., they're not causing it). Also, there are no other Track FX on any other track/buss in the signal flow downstream.

This is important because I'm bouncing each track to import into another DAW for another edit and I need an exact copy of the original, hence the careful mixdown of each original track.

Does anybody have any ideas?

[GrizzlyAK]

OK, so this is even more troubling than I could have imagined. After spending a full day troubleshooting this, I've narrowed the issue down to the following: The timing of Au's automation lanes do not correctly align with the track clip alignment when using Multitrack > Mix Down Session. In other words, what is rendered using the Mix Down process is different than what you hear,  see on the meters, and get when you Bounce a track using the exact same automation. This is a serious problem.

In my particular case, the desired state of an automation parameter (e.g., Send  5: Pan) needs to be set and established at least 185 ms prior to the start of a clip in order for Au to render the clip and the automation properly for a Mix Down. The track was originally pre-rendered, but I disabled that to ensure that wasn't introducing some unwanted delays (It made no difference).

Note that I was getting an echo in the Right channel from a mono file on a mono track being sent to Stereo track, even though the Send 5: Pan automation point was at 100% L visually well before the beginning of the clip.

Image 0 shows the original automation (Send 5: Pan is highlighted) relative to the clip in question. The Capture image in the previous post is what resulted from a Mixdown Session to File. As noted, the Pan is clearly NOT at 100% L in that mix down waveform.

Before doing these final tests, I eliminated the other two automation lanes related to other signal flows just to ensure they were not the cause (they weren't). In image 1, you will see where I moved the 100% L point back in time the approximately 185 ms I determined would give the proper (left only) mixdown, shown in image 2 (the cursor is ~185 ms prior to clip start). In image 3, I slid the Pan ramp to the right in time (cursor remains in original position throughout) to somewhere in-between the 185 ms point and the clip, which gives the mixdown shown in image 4. I slid the Pan ramp even further in time as shown in image 5, which you would think would still be 100% L, but which gave the mixdown shown in image 6. When I slid the Pan ramp to the position shown in image 7, I finally got a mixdown that was a Center Pan (both channels equal), as shown in image 8.

I am unsure why any position of the final automation state within 185 ms of the clip does not truly represent what is being downmixed. As I said, when I play it back, I hear only signal in the Left channel, and that is what is shown on the meters, and what is shown when the track is Bounced to a new track. Only the Mix Down has the timing defect. Depending on the scale of the project, this could be huge, and completely unexpected since it differs from what is seen in the DAW when mixing.

This has all but destroyed my trust in Audition's automation capabilities. I don't know if this defect still exists beyond CS6, as I no longer rely on Adobe products due to their Subscription-only model.

There was another anomaly I found that was just as disturbing. As mentioned, the mono track used above is Sent to a stereo Buss, which is (along with all the other individual 2.0 mix busses) Sent to a 2.0 Mix Buss before finally going to the 5.1 Master. When I solo the Track mentioned above, all of the tracks I just listed are soloed as well, as expected. If I Bounce to New Track each of these tracks in turn, I find that all but one correctly Bounces to a proper track with information in ONLY the Left channel. On the 2.0 Mix Buss, for some reason, a Bounce To New Track of that track produces a new track with the complete 2.0 mix (i.e., includes ALL the unsoloed tracks as well), even though the output flowing through it (that I hear and see on the Master, and on the 2.0 Mix Meters) is a Left Channel-Only signal.  There is something seriously wrong here. I suspect my routing here is more complex than the test team at Adobe has bothered with. I'll leave that exercise to the class... good luck.