TCE / X-Form Workflow Question

[Nirvana5253]

Hi All - Having a weird issue I hope is just operator error - hoping to confirm my workflow.

I'm making a D&E mix for a client and they want it delivered at 23.976. I made the D&E from my stems in the 24fps session. Everything was done at 24 and then converted to 23.976 in a separate session. I used the time shift plugin (audio suite) to go from 24 to 23.976.

The issue: When I take the 24fps full mix and null it against the 24fps D&E, all I get is music - perfect. Just what you would expect. When I take the 23.976 mix and null it against the 23.976 converted D&E, its WAY out of sync (up to 1200 samples) and I can hear everything - Dx, Mx, SFX. . I'm sure I'm not on varispeed in the time shift plug. I've tried it with x-form too and had similar results. X-Form got closer, but it still didn't null to just leave the music.

To confirm the 24->23.976 workflow, I took a new 24fps full mix and ran it through the time-shift plug in. I nulled this against the final delivery 23.976 full mix I sent - perfect null.

Any thoughts on this would be great. I've tried this in 10+CPTK and 11HD - same results.

[Nirvana5253]

Bump

Anyone have any thoughts on this? Thoroughly stumped. Will run it up the avid flag pole if no one here has any clue.

[reichman]

My only thought is that the algorithms inside X-Form and Serato PnT perform differently depending on the content. So a D&E mix will result in slightly different sounding audio than a full, for those D&E elements. The algorithm can only hear the total. Make sense?

There are bugs in the standard issue Time Shift plug-in. X-Form and PnT have always sounded better to me.

[Nirvana5253]

Thanks reichman - Totally understood that the algorithms are program-dependent.

What I don't understand is why at 24fps, when I null the full mix and D&E I get music onlybut at 23.976, I get full program and sync that drifts all over the place. Not sure if anyone from Avid is around here and could chime in. Normally I would deliver and be on my merry way but since I'm trying to match this to the already-delivered 23.976 full mix, I feel compelled to figure out what is going on / make sure I get it right. Thanks to anyone who might have some thoughts on this.

[minister]

The only way to reliably maintain sync when pulling down 24-->23.976 is to SRC. The pitch change is negligible.

[Nirvana5253]

Thanks minister. Coming to that conclusion myself. Seems using time-shift and x-form in this way are not best-case-scenarios. Will do the convert on bounce/import and run with it. Thanks all for your suggestions.

[Alexey Lukin]

Time/pitch change algorithms are not additive: a mix of stems will not scale as fine as individual stems. So, I wouldn't expect the null test to pass.

However, since your required amount of time stretch is only 0.1%, I would suggest using resampling (SRC) instead of time/pitch modification. Not only is resampling less destructive to the signal — it is also additive, which means that the null test will pass.

[Reckless Erik]

I agree here that SRC is the way to go since the difference in speed is so small. For those doing 25 to 24 frame conversions I have had very favourable results using SRC to change the speed and then using Time Factory II from zynaptiq to do the pitch conversion. When time is not a factor I have gone as far as only converting stems and bouncing out a new full mix. Works a treat.