SRC in SAW: a discovery (for me, anyway)

[Dave Labrecque]

I have a job where I'm doing lots and lots of manual plosive fixes (thank you Carl for that AUT file!). I'd been noticing some inconsistency between playback from the timeline with the de-pop automation and the build mix file from the same section. Seemed like there was some kind of "loss of sync" between the automation entries and the underlying audio -- but only during buildmixes. I came across a particularly bad area where there was bad waveform mishaping of bass frequencies (audible as kind of a click sound -- you know, like when a kick drum's gate or compressor is set to too fast an attack). Again, I figured it was a sync issue with the automation during the buildmix.

First I tried taking any UAD plugs off-line, thinking they were responsible. Didn't help -- I was actually getting the problem with no plugs -- just straight SAW mixer stuff. I finally discovered that the 44.1 KHz regions were being played in a 48 KHz session, and when I changed the MT sample rate to 44.1, the problem went away. Why it behaves differently on playback is a mystery to me.

Anyone come across this kind of thing? Bob -- want me to do some sample-accurate testing for ya?

[Bob L]

I thought roundoff errors might be causing an issue when doing SRC with automation entries... but I did some testing and did not see a problem.

I am still imagining plugins causing latency issues somewhere in your mix....

Here is the simple test I did... perhaps you can give me another that displays the problem.... this works fine.

A single track of transient material... place an automation entry at the exact peak of a rising waveform... I used a Shifted Fader drop to infinity from zero db... this created a zero slope cuttoff... build the mix at the same rate as the file... look at the result... perfect cuttoff at the peak of the transient... now I changed the MT rate and built with various SRC quality choices... in all cases the cuttoff was still perfect... I then used the buildmix dialog to change the final rate and again, all results were perfect.

Bob L

[Dave Labrecque]

I thought roundoff errors might be causing an issue when doing SRC with automation entries... but I did some testing and did not see a problem.

I am still imagining plugins causing latency issues somewhere in your mix....

Here is the simple test I did... perhaps you can give me another that displays the problem.... this works fine.

A single track of transient material... place an automation entry at the exact peak of a rising waveform... I used a Shifted Fader drop to infinity from zero db... this created a zero slope cuttoff... build the mix at the same rate as the file... look at the result... perfect cuttoff at the peak of the transient... now I changed the MT rate and built with various SRC quality choices... in all cases the cuttoff was still perfect... I then used the buildmix dialog to change the final rate and again, all results were perfect.

Bob L

OK, I'll take a look... when I can. Crazy week. I'm posting stuff during build mixes. :) That's all I can do right now. :eek:

[MMP]

If this EDL was started prior to the latest release, look for bad softedge boundries on the affected track prior to where you are hearing the automation mis-match. I recently had exactly the same issue where I initially thought the problem was being caused by SRC, but after further investigation, I found the cause to be an illegal softedge on the track.

So, try tabbing down the track and see first if this might be the cause.

Regards,

MM

[Dave Labrecque]

If this EDL was started prior to the latest release, look for bad softedge boundries on the affected track prior to where you are hearing the automation mis-match. I recently had exactly the same issue where I initially thought the problem was being caused by SRC, but after further investigation, I found the cause to be an illegal softedge on the track.

So, try tabbing down the track and see first if this might be the cause.

Regards,

MM

No softedges in this EDL. :confused: