OnMic — what's new

Current version: v1.22

Page 1 of 14
Next

v1.22 - better faint-mic handling

v1.22 - better faint-mic handling

- Fixed how OnMic reads Premiere's clip volume level, so a mic that was raised
  inside Premiere is analyzed closer to the level the editor actually hears.
- Improved bleed rejection for quiet mics that need a boost. When another mic is
  much stronger in the raw recording, OnMic is less likely to treat the boosted
  bleed as both people talking.
- The Log now always shows whether OnMic used Premiere source-channel labels or
  fell back to automatic WAV channel mapping, making wrong-channel reports much
  easier to diagnose from a support bundle.

v1.21 - respect Premiere source-channel routing

v1.21 - respect Premiere source-channel routing

- Fixed a multichannel WAV setup where Premiere showed timeline mic clips as
  specific source channels, such as Ch. 4 on A1 and Ch. 3 on A2, but OnMic could
  still analyze the file by track order or automatic channel order. That could
  make the edit cut to the opposite speaker even though the camera tags looked
  correct.
- OnMic now reads those Premiere source-channel labels when available, analyzes
  the exact channels shown on the timeline, and records the mapping in the Log
  and support bundle.

v1.20 - reliable Speakers and Cameras menus on Mac

v1.20 - reliable Speakers and Cameras menus on Mac

- Replaced the remaining native Speakers and Cameras count dropdowns with
  OnMic-rendered menus. They now open upward when the panel is short and scroll
  inside the panel instead of becoming invisible at Premiere's dock boundary.
- Saved profiles, automatic track counting, and analysis continue to use the
  same underlying count values.

v1.19 - split-recorder accuracy and speaker-tag menu fix

v1.19 - split-recorder accuracy and speaker-tag menu fix

- Fixed long sessions recorded across consecutive multichannel WAV files. Each
  speaker now keeps the correct isolated channel after the recorder starts its
  second, third, or later file instead of both speakers silently reading channel
  1. Mono replacement clips still use the safe channel-1 fallback.
- Speaker-tag menus now open above the panel's scrolling containers, choose the
  side of the field with available space, and keep mouse-wheel scrolling inside
  the menu. The choices no longer disappear below the card or action buttons.

v1.18 - reliable long-episode apply

v1.18 - reliable long-episode apply

- Fixed a Premiere focus race where creating the safety backup could switch the
  active timeline back to the backup during a long batched apply. OnMic now
  keeps a direct reference to the original sequence and reasserts both Premiere
  timeline views before every batch.
- Sequence IDs are compared consistently even when Premiere exposes the same ID
  as a number in one API and text in another.
- Apply now verifies every resulting camera-track visibility state before it
  reports success. A partial or conflicting assignment is rejected with clear
  instructions to restore the clean backup.

v1.17 - faster, more accurate speaker handoffs

v1.17 - faster, more accurate speaker handoffs

- Balanced pace now follows real speaker changes sooner instead of holding the
  previous camera through short answers and quick back-and-forth conversation.
- Tuned against a 53-minute producer reference edit: the new cadence closely
  matches its number of cuts while improving agreement
  on the active speaker.
- Low wide shots remain tied to genuine cross-talk and capped near 5%. Listener
  Reaction Shots Off remains a hard zero-reaction rule.

v1.16 - safer timeline edits and producer regression coverage

v1.16 - safer timeline edits and producer regression coverage

- OnMic now fingerprints the exact clip layout it analyzed. Apply, Trim Dead
  Air, markers, and reports stop if clips were moved, replaced, or retimed after
  analysis instead of applying an old plan to a changed sequence.
- Trim Dead Air verifies that every downstream audio and video track rippled by
  the same amount. If Premiere only moves some tracks, OnMic stops and returns
  to the clean backup sequence automatically.
- Added a producer-style 60-second regression corpus covering alternating
  speakers, listener-mic bleed, a silent third seat, Low wide mode, and Listener
  Reaction Shots Off.

v1.15 - no false listener shots from nodding or mic bleed

v1.15 - no false listener shots from nodding or mic bleed

- Fixed: Suggest camera tags could force a speaker assignment from extremely
  weak mouth-motion evidence, or fall back to whole-frame movement when no face
  was found. A listener nodding could therefore receive the speaker's camera tag
  and later look like a reaction shot even with Listener Reaction Shots Off.
- Weak/no-face camera suggestions are now left for manual tagging. Generic frame
  movement is never used to identify a speaker.
- Added a second audio safeguard against quiet listener-mic bleed being heavily
  boosted and mistaken for another speaker. Listener Reaction Shots Off remains
  a hard zero-reaction rule.
Page 1 of 14
Next

← Back to OnMic