Multicam Audio Rendering = Slow Exports, Consumes Invisible Disk Space
I'm currently cutting a feature doc with lots of multi-cam footage and sync'd audio. Whenever I export a sequence, Media Encoder (same thing happens if I export directly from Premiere, btw), has to "render" thousands of audio clips. From what I gather, it is converting every piece of audio in every channel of every multi-cam sequence, regardless of whether all channels of the Multicam are used in the edit, and regardless of how much or how little of a given piece of audio are used in the edit. For some reason, this is much slower when the Multicam audio has not been flattened. That is, if I leave the Multicam audio nested in the Multicam, the export is slower for the exact same audio with the same ins/outs. This means that for an hour-long sequence, it can easily take 6 hours to "render the audio", but then only about 30 or 45 mins to compress the video and compile an .mp4. Not only is this outrageous (I mean, it doesn't take 6 hours before I can play the sequence in Premiere), but to top it off, it fills up my boot drive with hundreds of GB of, I presume, cached audio files. The drive doesn't show in Finder as being out of space, but I get "out of space" errors. Then the finished export is missing audio in various spots, I presume again, b/c the maxed out boot drive had no where else to cache files. BTW, my boot drive has over 200 GB of free space. I have tried rendering the audio in Premiere before exporting, which takes forever (ie: the aforementioned 6 hrs), but even when I have "use previews" checked on export, the export doesn't use the rendered audio and does it all again from scratch. And of course after rendering audio in the sequence in premiere, any changes made mean that the entire sequence must be re-rendered, not just the portions affected by the change.
Can this audio problem be fixed? At very least, can Premiere/AME be intelligent enough to only render the ins/outs of the audio actually used in the sequence its exporting?
Can Preimere/AME use audio renders done in the sequence when it exports rather than re-rendering the entire sequence?
Can the audio renders done in the Sequence within Premiere be intelligent enough to not require re-rendering of the entire sequence when small changes are made?
Can there be a way to set the scratch disk to which AME/Premiere writes these cache files if/when it does have to render audio on export?
MacPro 6,1, Dual D700, 10.13, 64GB RAM, 1TB SSD internal. This issue has persisted in Premiere 2017 and 2018, and in OS 10.11, 10.12, & and 10.13.

-
madeinjapantonight commented
Look what happens when you try to export the waveform audio out of one of these types of multisequence nested timelines. I don't remember this happening before... obviously a bug of some sort.
-
Matvei Kurzukov commented
Same problem. Any ideas?