Is AI Speaker Tracking Actually Distracting Meeting Participants?

AI speaker tracking can distract remote participants when it cuts or pans too often in a lively, multi-person discussion, but it rarely does in presentations or meetings where one or two people do most of the talking. Most of the distraction comes from default settings, not from the technology: switching speed, framing size and the choice between tracking one speaker or framing the group can all be changed.

This post compares automatic speaker switching with fixed wide and group framing, and shows which settings vendors expose to calm things down. For a primer on what the features are, read Can AI-Powered Conference Cameras Finally Solve the Problem of Remote Participants Feeling Left Out? Our sibling posts ask whether AI cameras make meetings better or just busier and whether one USB video bar can handle 12 people.

Why can speaker tracking feel distracting?

A tracking camera has to make a decision every time someone new speaks: move to them, wait, or pull back to the group. Each decision is visible to the far end. In a two-person exchange that is helpful. In a six-person debate with short interjections, the picture can keep moving while the remote viewer is still working out who said what.

Vendors build in delays for exactly this reason. Logitech's admin documentation for RightSight 2 on Rally Bar and Rally Bar Mini says that when two people speak, Speaker View first frames both and "will eventually frame only the new speaker unless interrupted", and after a period of silence it reframes the whole room. That is a deliberate trade-off between responsiveness and stability, and it is tunable.

There is also a technical source of distraction. HP notes in its Camera Control App admin guide that pan, tilt and zoom on the Poly Studio R30, Studio USB and Studio E70 can look choppy because these cameras use digital zoom, and recommends recalling a preset instead. Electronic zoom is cropping a fixed sensor, so aggressive moves can look less smooth than a motorised camera.

Speaker tracking, group framing or fixed wide view: what is the difference?

Most current video bars offer at least three behaviours, and newer ones add a fourth: several people shown at once in separate tiles.

Examples of the tile approach include Microsoft IntelliFrame on Teams Rooms Pro, Yealink IntelliFocus on the Yealink MeetingBar A40 (up to four portraits, according to the datasheet), Jabra Dynamic Composition on the Jabra PanaCast 50, and Logitech Grid View in RightSight 2.

Is a multi-person layout a calmer compromise?

Often, yes, because the layout stays put while the content of each tile changes. Microsoft's IntelliFrame support page describes four larger frames: the active speaker in one, and the three most recent speakers in the others, with a room panorama below by default. Remote users can right-click the tile and choose to hide the panorama; people in the room can change the camera mode from the console.

Microsoft also publishes its limits, which are useful for judging whether a room suits the feature: up to 12 people in the room, no more than 64 meeting participants, people at least a metre apart and within 5 metres of the camera. Microsoft warns the camera "may identify inanimate objects as people" when there is movement, shadows or wind. In practice that means a glass-walled room facing a busy corridor is a poor candidate.

Which settings make speaker tracking less distracting?

The controls differ by vendor, but three dials appear repeatedly in the documentation.

Switching and framing speed

Poly's Studio USB user guide lists a Tracking Speed setting of Slow, Normal (default) or Fast. Logitech's RightSight 2 offers separate Speaker Detection and Framing speed controls. Logitech states that Speaker View frames the active speaker in roughly 10 to 20 seconds at Faster, 20 to 25 at Default and 25 to 30 at Slower, while Group View reframes in roughly 5 to 7, 7 to 10 and 10 to 12 seconds. For a room that feels jumpy, the slower setting is the first thing to try.

Framing size and maximum zoom

The same Poly guide offers Tight, Medium (default) and Wide framing sizes, plus a maximum zoom cap of 2x, 3x or 4x in tracking modes. A medium or wide frame keeps neighbours and reactions in shot, which reduces the sense of being cut away from the conversation.

Cut versus pan

Poly lets admins choose Auto Pan, which moves smoothly between speakers or groups, or Cut, which switches directly. Which is less distracting is a matter of taste; a clean cut avoids a slow sweep across the table, while a pan keeps viewers oriented. Try both with your regular remote attendees.

Jabra takes a different route: its product page says the PanaCast 50's 180° view can be narrowed in Jabra Direct and Jabra Xpress, which helps when the panorama picks up empty corners or a glass wall.

Which mode should each type of room use?

Based on the vendor guidance above, a sensible starting point looks like this:

• Huddle rooms (two to four people): group framing or a fixed wide view. Speaker tracking adds movement without adding much information.
• General meeting rooms: group framing by default, with multi-person tiles where the platform supports them.
• Training and presentation rooms: speaker or presenter tracking. Logitech recommends Speaker View for rooms where typically only one or two people speak, and Poly's guide suggests keeping the bar within 3 m of a presenter for presenter mode.
• Debate-heavy rooms (project reviews, stand-ups): tiles or group framing, with speaker tracking off.

HP's Camera Control App admin guide lists Group, People (up to six participants) and Speaker framing on most supported Poly cameras, with Presenter framing on the Studio E60, so a single camera can often cover several of these room types by changing one setting.

Which video bars give you the most control?

Prices as listed on 25 Oct 2026.

• HP Poly Studio USB Video Bar (Price on request): group and speaker framing with tracking speed, framing size, zoom cap and cut or pan options, per Poly's user guide.
• HP Poly Studio R30 USB Video Bar (Price on request): HP lists speaker, group and people framing plus presenter tracking for small rooms.
• Logitech Rally Bar Mini (Price on request): RightSight 2 Speaker, Group and Grid views with adjustable speeds.
• Jabra PanaCast 50 (S$1,575.00): Virtual Director, Dynamic Composition and an adjustable field of view.
• Yealink UVC40 E2 (S$645.00): auto framing and speaker tracking on a 120° USB bar for small and huddle rooms.

Compare the Poly USB line-up in Poly Studio R30 vs V12 and the Yealink range in Yealink MeetingBar A10 vs A25 vs A40 vs A50. Framing behaviour also changes with firmware, so keep updates controlled, as discussed in The Next Big Software Update Could Break Your Meeting Room.

If remote attendees complain about the camera, slow it down and widen the frame before you replace anything.

What to do next: a tuning checklist

1. Ask the far end. Remote regulars notice camera movement more than the room does.
2. Match the mode to the room type, using the list above, and write it down as the room standard.
3. Slow switching first. Set tracking or framing speed to its slower option.
4. Widen the frame. Choose medium or wide framing, or cap the maximum zoom.
5. Remove false triggers. Narrow the field of view or use boundary features in glass-walled rooms.
6. Check audio. Tracking that relies on sound direction works poorly if microphones struggle; see why an expensive meeting room can still sound terrible.
7. Lock settings centrally through the vendor's management tool so they survive resets.

Frequently asked questions

Is AI speaker tracking distracting? It can be in fast multi-person discussions, where the camera keeps switching. In presentations or meetings with one or two main speakers it usually helps remote viewers follow the conversation.

What is the difference between speaker tracking and group framing? Speaker tracking shows a close-up of whoever is talking. Group framing fits everyone present into one shot and only reframes when people arrive or leave.

Can I slow down speaker tracking? On many bars, yes. Poly's Studio USB has Slow, Normal and Fast tracking speeds, and Logitech's RightSight 2 has adjustable speaker detection and framing speeds.

Is a fixed wide view better than AI framing? In a small huddle room or a fast debate, often yes. In medium rooms the far-end faces become too small, so group framing or multi-person tiles usually work better.

What does Microsoft IntelliFrame do differently? It shows in-room participants in separate frames, with the active and recent speakers in larger tiles and a room panorama below. Microsoft states it needs Teams Rooms Pro and a supported camera.