VIBI: AI Sound Eraser is a UXP panel that works inside Premiere Pro to separate a clip's audio into individual components. You can load a local file or pull a clip from the Project panel or Timeline, with a waveform preview to confirm the source. One click splits the track into per-speaker stems—Speaker 1, Speaker 2, and so on—plus a Background stem, using about 1 credit per minute.
From there, a stem mixer lets you toggle each stem on or off and ride its level with a per-stem dB slider. A timecoded transcript shows who said what and when, and you can rename or add speakers, tap any line to reassign it, then regenerate the audio to match your edits. The finished mix exports to your Project or Timeline as a .wav, with no round-trip to another app.
The panel is aimed at interviews and podcasts, documentary and vlog footage that can't be reshot, and removing music from reference cuts before your own score goes in. A companion VIBI iPhone app handles simpler cleanups, and one account shares credits across both.