The interview is going well until two people answer at once. If both voices went into one microphone, lowering one fader will not separate them. Adobe has moved Separate Crosstalk into Premiere 27.0: a cloud tool intended to pull apart that overlap, giving an editor another option before asking for a re-record.
Based on 5 reviewed sources · See sources
A targeted repair, with a listening check
Adobe’s production announcement describes separate tracks for two speakers and the background, with the original audio preserved and disabled. It processes one clip at a time, from one second to ten minutes. This is a repair for overlapping speech, rather than a promise to identify every speaker throughout a project.
An earlier beta test gives a useful reason to keep the selection small. Adobe Community Expert Stan Jones described a synthetic 94-second stereo recording on Windows 11 and Premiere Beta 26.5.0.77. He reported useful separation but incorrect speaker assignment in some non-overlapping passages. That is an attributed beta observation; we have not tested the production release.
Adobe’s broader Premiere update announcement includes AI Audio Tools; the linked production documentation supplies Separate Crosstalk’s exact scope.
View the original post on X ↗Trim the moment, check the estimate
The processing guide prices the feature at one generative credit per processed second. By default it adds two seconds of surrounding audio on each side, which also count; the estimate appears before processing. A ten-second selection with both full handles available would therefore mean fourteen processed seconds, or fourteen credits. That is a calculation from the documented rule, not a measured invoice.
Adobe’s cloud-voice guidance requires sign-in, an internet connection and consent confirming the necessary rights for the people recorded. It says voice embeddings—numerical representations of voice characteristics that may involve biometric information—are used to provide the feature, not retained beyond that purpose or used to train its AI. Uploads from Illinois and Texas are unavailable. Editors handling interviews should check that processing permission before uploading, then compare the separated result with the original for missing words or altered voices.
Sources
- Fix overlapping interview dialogue with Separate Crosstalk in Adobe Premiere 27.0
Adobe / Kevin Monahan · Announcement ·
- Separate overlapping speakers
Adobe Help · Documentation ·
- About cloud-based AI voice features
Adobe Help · Documentation ·
- Earlier beta announcement and Stan Jones’s test
Adobe Community · Event recap
- Premiere update announcement
Adobe Video on X · Announcement ·
Written with AI assistance from the sources above. Analysis and practical implications are our interpretation. Our editorial approach.



