Repository navigation
Land the spatial rendering fixes (PR #164) on master - #165
Merged
Merged
Conversation
Browsers stop requestAnimationFrame for a hidden or minimised tab while audio keeps playing, so a glide froze part-way until the tab was shown again. The tweener now takes a visibility source (the document by default, injectable so it stays testable without a DOM): a tween asked for while hidden snaps to its target, and glides in flight when the document becomes hidden land on their targets at once. Nothing is left in flight, so nothing moves when the tab is shown again. Adds VectorTweener.finishAll(), and an idle tweener now holds neither a pending frame nor a visibility listener.
Sound positions glide on MediaService's own VectorTweener, a separate instance from the Client.Spatial handler's, so the handler's cancel on Client.Spatial.Scene never reached them: a sound kept gliding across a room change, between two rooms' unrelated coordinates. The Scene handler now calls MediaService.sceneChanged(). Glides in flight land on their targets at once (the last position the server gave, never somewhere part-way), and the first position each sound receives afterwards is placed directly instead of gliding from a position in the previous room's coordinates.
Two handlers wrote the Cacophony listener position. ListenerPosition glided it every frame at the previously stored velocity; the listener's own EntityMove wrote the target outright when its entity glide finished, at the new velocity. The server sends the mover both for one step, so with different speeds the listener jumped to the target and then back onto the slower glide. Both now feed one glide on the listener tween key, which is the only writer while the listener moves. The second message for a step retargets that glide instead of starting a competing one, and it runs at the speed the step itself carries. An EntityMove for the listener that arrives alone still moves the listener, now per frame.
A LiveKit speaker with no position was placed at [0, 0, 0]: the centre of the listener's current room at floor height. A phone-call partner, or a speaker still connected from the room just left, was heard as a point there, getting louder and swinging around as the listener walked. A voice is now positional only while its speaker is an entity in the current scene. The bridge passes null for anyone else, and MediaVoices plays that voice as a stereo source at centre pan: it has no 3D panner, so no position, no distance model and no direction. Cacophony fixes a stream sound's panning mode at creation, so when the entity appears or leaves, the voice's sound is rebuilt in the other mode on the same track, keeping the named route it had or still wanted. A voice that becomes positional is placed before it plays. The test that asserted origin placement is rewritten to the new rule.
A Play for a playing key recreated its voice whenever start differed, because the segment comparison included start. start is the join cursor, not part of the region. The server re-Plays every audible sound at its current cursor to a listener who enters a room, so every continuing sound (music, a radio, a fire) was cut and restarted at every doorway. Region identity is now the repeat window, loopStart..finish, as the wire contract says: same key, source and region keeps the voice, and an explicit start seeks it. The MOO sends loopStart only for a sound that repeats, so a single pass without it is kept when the voice's region reaches back to the new cursor; a repeating sound without loopStart still has a repeat window that follows start. On a kept voice the seek is skipped when the playhead is already within MEDIA_SEEK_TOLERANCE_MS (150 ms) of the cursor, measured across the loop point for a looping voice, so a re-Play at the cursor the voice is at is seamless. An engine that reports no playhead seeks as before, and continue: false still forces a new voice.
A source's panning mode is fixed when Cacophony creates it, and nothing handled a later change of is3d. The server sends is3d: false with no position for a sound heard through a doorway that has no known point, and is3d: true with a position once the listener is in its room. A kept voice stayed in the wrong mode: still HRTF at its last coordinates, or stereo and attenuated by distance with no direction. Keeping voices across re-Plays makes this the common path at every doorway. A Play or an Update that changes is3d for a playing key now builds the same source and region in the new mode. The old voice plays on while the replacement loads; at the swap the new voice starts at the old one's playhead, with no second fade-in, unless the Play names a cursor further than the seek tolerance away. Volume, gain, pitch, spatial profile, loop count, occlusion and routing carry over: an Update's fields are now folded into the payload that describes the sound, so a rebuild (this one, or a moved segment) replays the state in effect now rather than the original Play's. A Stop or a newer Play during the load wins, and the replaced voice is released on every exit. Ambisonic sounds are not rebuilt here, and a preloaded sound that was never Played is left as it was.
Fix six spatial-audio rendering defects
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
1 participant
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.
PR #164 was stacked on
feat/media-occlusionand was merged into that branch 18 seconds after PR #163 had merged the branch into master, so its six commits never reached master.This PR carries exactly those commits (already reviewed and merged as #164) to master:
No new changes. All tests mock the audio engine; none of this has been listened to in a browser.