feat(audio-rooms): pre-roll buffering overlay on speaker avatars

Tracks the window between SUBSCRIBE_OK and the first decoded audio
frame (typically 0.5-2s on a fresh subscription) so the user can
see audio is on its way rather than wonder why a "live" speaker
is silent.

Wire-up:
  * AudioRoomUiState gains `connectingSpeakers: ImmutableSet<String>`,
    a set-once-per-subscription field.
  * VM enters the set on `slot.attach(...)` (right after SUBSCRIBE_OK
    succeeds), exits on the first `onSpeakerActivity` callback (first
    decoded packet), and clears on speaker removal. Set membership
    survives subsequent silence — once a speaker has delivered a
    frame, we know the pipeline works.
  * ParticipantsGrid renders a small CircularProgressIndicator
    overlaid on the avatar (sized smaller than the picture so the
    user stays recognisable underneath).

Audience-side avatars don't get the overlay — they have no MoQ
subscription. Only on-stage speakers in the connectingSpeakers set
show it.

amethyst/src/main/java/com/vitorpamplona/amethyst/ui/screen/loggedIn/audiorooms/room/AudioRoomFullScreen.kt

@@ -206,6 +206,7 @@ internal fun AudioRoomFullScreen(
                 onStageLabel = stringRes(R.string.audio_room_stage),
                 audienceLabel = stringRes(R.string.audio_room_audience),
                 reactionsByPubkey = reactionsByPubkey,
+                connectingSpeakers = ui.connectingSpeakers,
                 onLongPressParticipant = onLongPressParticipant,
             )
             val speakerCatalogs by viewModel.speakerCatalogs.collectAsState()

amethyst/src/main/java/com/vitorpamplona/amethyst/ui/screen/loggedIn/audiorooms/room/ParticipantsGrid.kt

@@ -27,6 +27,7 @@ import androidx.compose.foundation.layout.Spacer
 import androidx.compose.foundation.layout.fillMaxWidth
 import androidx.compose.foundation.layout.height
 import androidx.compose.foundation.layout.padding
+import androidx.compose.foundation.layout.size
 import androidx.compose.foundation.layout.width
 import androidx.compose.foundation.lazy.grid.GridCells
 import androidx.compose.foundation.lazy.grid.LazyHorizontalGrid
@@ -72,6 +73,7 @@ internal fun ParticipantsGrid(
     audienceLabel: String,
     modifier: Modifier = Modifier,
     reactionsByPubkey: Map<String, List<RoomReaction>> = emptyMap(),
+    connectingSpeakers: ImmutableSet<String> = kotlinx.collections.immutable.persistentSetOf(),
     onLongPressParticipant: ((String) -> Unit)? = null,
 ) {
     Column(modifier = modifier.fillMaxWidth()) {
@@ -81,6 +83,7 @@ internal fun ParticipantsGrid(
                 members = grid.onStage,
                 avatarSize = Size40dp,
                 speakingNow = speakingNow,
+                connectingSpeakers = connectingSpeakers,
                 accountViewModel = accountViewModel,
                 reactionsByPubkey = reactionsByPubkey,
                 onLongPressParticipant = onLongPressParticipant,
@@ -92,10 +95,11 @@ internal fun ParticipantsGrid(
                 title = audienceLabel,
                 members = grid.audience,
                 avatarSize = Size35dp,
-                // Audience rows don't get the speaking-ring — only
-                // members with a live broadcast track do. Pass an
-                // empty set rather than thread a per-section bool.
+                // Audience rows don't get the speaking-ring or
+                // buffering overlay — only members with a live
+                // broadcast subscription do.
                 speakingNow = kotlinx.collections.immutable.persistentSetOf(),
+                connectingSpeakers = kotlinx.collections.immutable.persistentSetOf(),
                 accountViewModel = accountViewModel,
                 reactionsByPubkey = emptyMap(),
                 onLongPressParticipant = onLongPressParticipant,
@@ -110,6 +114,7 @@ private fun ParticipantsSection(
     members: List<RoomMember>,
     avatarSize: androidx.compose.ui.unit.Dp,
     speakingNow: ImmutableSet<String>,
+    connectingSpeakers: ImmutableSet<String>,
     accountViewModel: AccountViewModel,
     reactionsByPubkey: Map<String, List<RoomReaction>>,
     onLongPressParticipant: ((String) -> Unit)?,
@@ -145,17 +150,33 @@ private fun ParticipantsSection(
                         com.vitorpamplona.amethyst.model.LocalCache
                             .getOrCreateUser(member.pubkey)
                     }
+                val isConnecting = member.pubkey in connectingSpeakers
                 Column(
                     horizontalAlignment = Alignment.CenterHorizontally,
                     modifier = Modifier.width(avatarSize + 16.dp),
                 ) {
-                    ClickableUserPicture(
-                        baseUserHex = member.pubkey,
-                        size = avatarSize,
-                        accountViewModel = accountViewModel,
-                        modifier = avatarModifier,
-                        onLongClick = onLongPressParticipant?.let { cb -> { hex -> cb(hex) } },
-                    )
+                    androidx.compose.foundation.layout.Box(contentAlignment = Alignment.Center) {
+                        ClickableUserPicture(
+                            baseUserHex = member.pubkey,
+                            size = avatarSize,
+                            accountViewModel = accountViewModel,
+                            modifier = avatarModifier,
+                            onLongClick = onLongPressParticipant?.let { cb -> { hex -> cb(hex) } },
+                        )
+                        if (isConnecting) {
+                            // Pre-roll buffering overlay — visible
+                            // between SUBSCRIBE_OK and the first
+                            // decoded frame (typically 0.5-2 s on a
+                            // fresh subscription). Sized smaller than
+                            // the avatar so the user picture stays
+                            // recognisable underneath.
+                            androidx.compose.material3.CircularProgressIndicator(
+                                modifier = Modifier.size(avatarSize - 8.dp),
+                                strokeWidth = 2.dp,
+                                color = MaterialTheme.colorScheme.primary,
+                            )
+                        }
+                    }
                     com.vitorpamplona.amethyst.ui.note.UsernameDisplay(
                         baseUser = user,
                         weight = Modifier.fillMaxWidth(),

commons/src/commonMain/kotlin/com/vitorpamplona/amethyst/commons/viewmodels/AudioRoomViewModel.kt

@@ -628,6 +628,9 @@ class AudioRoomViewModel(
         if (_uiState.value.speakingNow.contains(slot.pubkey)) {
             _uiState.update { it.copy(speakingNow = (it.speakingNow - slot.pubkey).toPersistentSet()) }
         }
+        if (_uiState.value.connectingSpeakers.contains(slot.pubkey)) {
+            _uiState.update { it.copy(connectingSpeakers = (it.connectingSpeakers - slot.pubkey).toPersistentSet()) }
+        }
         if (_speakerCatalogs.value.containsKey(slot.pubkey)) {
             _speakerCatalogs.update { it - slot.pubkey }
         }
@@ -710,6 +713,12 @@ class AudioRoomViewModel(
                 roomPlayer.play(instrumented, onError = { /* swallow per-packet decoder errors */ })
                 slot.attach(handle, roomPlayer, player)
                 publishActiveSpeakers()
+                // Enter the buffering window — UI renders a spinner
+                // overlay until the first audio frame triggers
+                // `onSpeakerActivity` and clears the entry.
+                _uiState.update {
+                    it.copy(connectingSpeakers = (it.connectingSpeakers + pubkey).toPersistentSet())
+                }
             } catch (t: Throwable) {
                 // Either CancellationException (scope cancelled mid-construction)
                 // or an unexpected throw — release the half-built pipeline and
@@ -792,6 +801,11 @@ class AudioRoomViewModel(
     private fun onSpeakerActivity(pubkey: String) {
         if (closed) return
         speakingExpiryJobs[pubkey]?.cancel()
+        // First frame for this subscription — clear the buffering
+        // overlay. Subsequent frames are no-ops here.
+        if (_uiState.value.connectingSpeakers.contains(pubkey)) {
+            _uiState.update { it.copy(connectingSpeakers = (it.connectingSpeakers - pubkey).toPersistentSet()) }
+        }
         if (!_uiState.value.speakingNow.contains(pubkey)) {
             _uiState.update { it.copy(speakingNow = (it.speakingNow + pubkey).toPersistentSet()) }
         }
@@ -914,6 +928,18 @@ data class AudioRoomUiState(
     val activeSpeakers: ImmutableSet<String> = persistentSetOf(),
     /** Pubkeys whose audio track delivered an object in the last [SPEAKING_TIMEOUT_MS]. */
     val speakingNow: ImmutableSet<String> = persistentSetOf(),
+    /**
+     * Pubkeys we have an open subscription for but no audio frame has
+     * arrived yet — the pre-roll window between SUBSCRIBE_OK and the
+     * first decoded packet. Typically 0.5-2 s on a fresh join. The UI
+     * shows a buffering overlay here so the user knows audio is on
+     * its way (vs the speaker sitting silent on stage).
+     *
+     * Membership is set-once-per-subscription: once a frame arrives
+     * the pubkey moves out of this set and stays out, even if the
+     * speaker subsequently goes quiet. Cleared on speaker removal.
+     */
+    val connectingSpeakers: ImmutableSet<String> = persistentSetOf(),
     /** Speaker / publisher state — only relevant when [AudioRoomViewModel.canBroadcast]. */
     val broadcast: BroadcastUiState = BroadcastUiState.Idle,
     /**