feat(call): multi speaker view - #18966
Open
DorraJaouad wants to merge 8 commits into
Open
Conversation
DorraJaouad
added
enhancement
feature: frontend 🖌️
Speaker view will be able to promote several active speakers at once instead of a single one. Add the state driving it, persisted next to the grid preference so it survives rejoining a call. The flag is only meaningful while speaker view is active, so switching between grid and speaker view leaves it untouched. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
A participant speaking without interruption for 3 seconds becomes active, and stops being active after a minute of silence, so that short interjections never disturb the view and listeners make room on their own. The active speakers are kept in the order they became active and only the first four are promoted. A fifth speaker is therefore queued rather than dropped: it slides in as soon as one of the speakers ahead is demoted, which keeps the promoted tiles stable instead of letting a single word evict someone mid-sentence. Participants that leave free their spot immediately, as the speaking flag is not guaranteed to be reset on disconnection. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
Whether a slot has to be reserved for the local video is a property of the grid being laid out, not something the layout can derive on its own. Move the decision to the caller so that grids other than the participant one can be measured with the same composable. No functional change: the videos grid passes the very condition the composable used to compute. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
The component lays out the handful of promoted speakers in the main area, reusing the grid layout helpers: the number of columns and rows follows the available space and the target tile aspect ratio, and the half column placement centers a row that leaves an odd number of columns empty, so three speakers show one centered tile below two. It is deliberately not the videos grid: no pagination, no local video and no tile ordering, as the caller already hands over the tiles to show in the order they should appear. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
In multi-speaker mode the promoted speakers are laid out in the main area instead of the single last speaker, and are taken out of the stripe: a tile is only ever in one place, so it reads as having moved up rather than as being duplicated. A screen share or a manually selected video still takes the whole main area, and the single speaker layout is kept until a second speaker becomes active, so the view only splits once there really is a conversation. Speakers shown in the main area now also get the video quality a promoted participant gets, otherwise all but one of the tiles would be upscaled from the lowest simulcast layer. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
The layout button used to toggle between the grid and speaker view, which leaves no room for a third option. Turn it into a menu listing the three layouts, so the current one is also readable at a glance instead of being implied by the icon. Picking the grid leaves the speaker view preference alone, so returning to it restores the layout that was in use. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
A promoted tile is rendered by the stripe before the change and by the speakers grid after it, so it cannot simply be transitioned: it is a new element in a new place, and it used to appear in its slot out of nowhere. Put the new tile back over the old one with a transform and release it, so that it reads as a single tile travelling from the stripe into its slot, and back into the stripe once its speaker is demoted. The animation is skipped for users who asked not to be shown motion. Assisted-by: ClaudeCode:claude-opus-5 Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
Signed-off-by: Dorra Jaouad <dorra.jaoued7@gmail.com>
DorraJaouad
force-pushed
the
feat/noid/multi-speaker-view
branch
from
August 10, 2026 12:17
aa1d47d to
a69cca7
Compare
Antreesy
reviewed
Aug 11, 2026
| :token="token" | ||
| :isOverlap="showFullPage" | ||
| :callParticipantModels="callParticipantModels" | ||
| :callParticipantModels="stripeParticipantModels" |
Contributor
There was a problem hiding this comment.
VideosGrid can now receive only one model (because the rest are filtered) and give false positive on isLessThanTwoVideos computed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
☑️ Resolves
It is a big PR bc it is a giant feature. The logic behind multi speaker is to use automation based on speaking. The demotion happens after 1 min of silence. It is max 4 speakers.
AI (if applicable)
I prepared a simulation in the last commit to help test, you can alter it to test edge cases of course (I included the case of 5 to 6 speakers). This is what is expected
🖌️ UI Checklist
🖼️ Screenshots / Screencasts
Screen.Recording.2026-08-09.at.18.37.52.mov
🏁 Checklist