6 min read
Exporting audio from Premiere Pro: mixdown, multitrack and the settings people miss
Premiere defaults to a stereo AAC mixdown. That is the wrong file for almost every sound handoff.
Premiere Pro exports audio well once you stop using the presets. The defaults are built for uploading to the web, not for handing a project to a sound department, and the difference shows up as a 320 kbps stereo file where a 24 bit multichannel WAV was expected.
The correct export recipe
- File, Export, Media, then set Format to Waveform Audio, which is WAV.
- Uncheck Export Video so nothing is rendered twice.
- Audio codec: Uncompressed. Sample rate: 48000 Hz. Sample size: 24 bit.
- Channels: stereo for a mixdown, multichannel when the sequence is 5.1 or when you need discrete tracks.
Getting separate tracks instead of a mixdown
Premiere does not have a true one file per track button. The reliable route is a multichannel sequence: set the sequence audio master to multichannel, assign each source track to its own output channel, then export a multichannel WAV. You get one interleaved file with every track intact.
From there, split that file into mono stems. The channel order follows your output assignments, so write it down before you export. The audio inspector on this site shows the resulting channel count and any embedded names, which is a fast way to check you assigned what you thought you did.
Send to Audition, or hand over an OMF
For a small job, Edit in Adobe Audition creates a session with your clips intact. For anyone working in Pro Tools or Nuendo, export AAF instead, with handles of at least two seconds. AAF carries clip boundaries and levels; a flat mixdown carries nothing.
Why AAC keeps sneaking in
Every H.264 preset uses AAC audio. If you export a review copy and then pull the audio out of it for the mix, you are mixing from a lossy file. Extract audio only from PCM masters, and keep the review encode strictly for review.
Quick check before delivery
Drop the exported WAV into the audio inspector, confirm 48 kHz and 24 bit, then listen to the first and last seconds. If the file is a few frames short, the export range was set to the work area rather than the whole sequence.