fix(voice): route media on evidence and switch devices without dropping the call

Outgoing voice was gated on the observer's roster copy of the remote user's
voice state, which is signaling gossip. The signal server broadcasts
`user_left` for any socket it declares dead, so a suspended laptop or a flaky
hop wiped that copy and the observer detached its microphone from a peer that
never left the channel - a silent member with no way back through the UI.

`decideVoicePathRouting` now closes a path only on positive evidence: we left
voice, the peer itself reported another channel or none, or the connection is
gone. Missing gossip holds an established path instead. Opening still needs
confirmation, so a guess never starts sending; the same rule gates playback,
camera video, and the microphone a new connection puts in its first offer.
Peers are also asked for their voice state when a connection or data channel
comes up, so a rebuilt path re-confirms itself.

Alongside it, the microphone can be switched mid-call: capture moves to a
device service and rules, the live track is swapped with `replaceTrack` so
the session is never renegotiated, and the speaking indicator follows the new
stream.
This commit is contained in:
2026-08-14 03:19:29 +02:00
parent a83f5aa750
commit 92c2f578e2
33 changed files with 3149 additions and 317 deletions
@@ -7,7 +7,6 @@ import { Store } from '@ngrx/store';
import { STORAGE_KEY_USER_VOLUMES } from '../../../../core/constants';
import { jsonStorage } from '../../../../infrastructure/persistence/json-storage.service';
import { ScreenShareFacade } from '../../../../domains/screen-share';
import { User } from '../../../../shared-kernel';
import { selectAllUsers, selectCurrentUser } from '../../../../store/users/users.selectors';
import { VoiceConnectionFacade } from '../facades/voice-connection.facade';
@@ -331,7 +330,7 @@ export class VoicePlaybackService {
if (!pipeline)
return;
if (this.deafened || this.captureEchoSuppressed || this.isUserMuted(peerId) || !this.isPeerInCurrentVoiceRoom(peerId)) {
if (this.deafened || this.captureEchoSuppressed || this.isUserMuted(peerId) || !this.mayHearPeer(peerId)) {
pipeline.gainNode.gain.value = 0;
return;
}
@@ -346,22 +345,17 @@ export class VoicePlaybackService {
this.peerPipelines.forEach((_pipeline, peerId) => this.applyGain(peerId));
}
private isPeerInCurrentVoiceRoom(peerId: string): boolean {
const localVoiceState = this.currentUser()?.voiceState;
if (!localVoiceState?.isConnected || !localVoiceState.roomId || !localVoiceState.serverId) {
return false;
}
const remoteVoiceState = this.findUserForPeer(peerId)?.voiceState;
return !!remoteVoiceState?.isConnected
&& remoteVoiceState.roomId === localVoiceState.roomId
&& remoteVoiceState.serverId === localVoiceState.serverId;
}
private findUserForPeer(peerId: string): User | undefined {
return this.allUsers().find((user) => user.id === peerId || user.oderId === peerId || user.peerId === peerId);
/**
* Whether this peer's audio may be audible.
*
* The roster is gossip: the signal server broadcasts `user_left` for any socket it
* declares dead, which used to mute a peer still sitting in our channel. Playback
* therefore asks the voice session, which weighs the roster against what the peer
* itself reported over its data channel, and holds a stream we already receive when
* neither confirms nor denies.
*/
private mayHearPeer(peerId: string): boolean {
return this.voiceConnection.mayHearPeerVoice(peerId, this.rawRemoteStreams.has(peerId));
}
private syncOutgoingVoiceRouting(): void {