804a97b0a9f867816c409ce8ba037e93fd78c195
10
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  adminandClaude Opus 5.5
|
804a97b0a9 |
Hand out the length of a played sound from Play sound
Wired into the time of Set sound FX parameters, a change then spans exactly one run of the clip, so a fade or a filter sweep ends with its last sample. - Play sound has a second output, length, in seconds. It is 0 where nothing was played or the sound has not loaded yet. - The AudioSink gives the length of a file, the SoundLibrary takes it from the decoded buffer. A pitch the voice starts on keeps it (#312). - A pitch changed while the voice plays still changes its length, so a pitch sweep over length does not end with the sound. A pitch that keeps the length in real time is #317. Tests: - soundfx, sounds, keep-on-scene-load, node-catalog, sound-library: 462 tests green. - Counter-check: with length always 0, the catalog case and the new soundfx case fail. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
e7a360db73 |
Offer the easings of an animation key on Lerp and sound FX ramps
A timed Set sound FX parameters ran only linearly, and a Lerp had four quadratic eases of its own. An animation key already had the full set. Now all three offer the same choice: Linear, Step, the ten curves from Sine to Bounce with In, Out or In-out, and a drawn Curve asset. - easingParams in src/core/graph/nodes/easing.ts gives a node the fields Easing, Direction and Curve from the key's own lists, and easingOfParams reads them back as a key interpolation. - There is no migration. A Lerp saved with the old in, out or inOut runs linear until its easing is picked again. - Set sound FX parameters hands the easing to the player. Linear and Step stay ramps of the audio clock, every other easing goes to setValueCurveAtTime as 200 samples a second, in semitones and dB. - Step now reaches its end at t 1. A key never asked for that, it reads the next key itself, so animations play as before. - The curve-in-use count and the warning for a gone curve cover the sound node too. Tests: - node-catalog: Lerp with cubic in, quad out and in-out, a curve without a direction, step, an unknown easing as linear. The sound node hands on a curve, step and a gone curve as linear. - sound-library: an eased change becomes a curve in semitones and decibels, a step a jump at the end. - node-catalog and curve after dropping the old values: 399, green. - e2e 6, 34, 53 on melektaus-atx: ALL PASSED, 198 checks. In Chrome a 2 s Quad in from depth 7 to -5 read speed 0.864 early and 0.5 at the end, a 1 s Step held 1 and then jumped to 0.5. - Counter-check: with curves sent as linear ramps, the curve test of the sound library fails. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
501a1524ed |
Ramp sound FX parameters over a time on the audio clock
A pitch sweep counted down in On update or On fixed update runs in game time, while the sound runs on the clock of the audio device. A short sound magnifies any offset between the two, so the sweep came out different on every play and most of all on the first. - Set sound FX parameters has a new input time in seconds. The values run evenly to their target in that time, scheduled on the audio clock with linearRampToValueAtTime or exponentialRampToValueAtTime. - Evenly in their own unit: a pitch in semitones and a gain in decibels, an EQ band and a Mix share linearly in their value. - A time of 0 or below keeps the quick glide of about 0.1 s. - Each change starts from where the voice is now. Whatever was still planned is dropped, so a change never queues behind an older one. Firefox lacks cancelAndHoldAtTime, there the value is held by hand. - A voice still waiting on its bytes starts a timed change with the sound, so the whole sweep is heard. Tests: - node-catalog: time is handed on, a time below 0 is a quick glide. - sound-library: semitones and decibels ramp in their own steps, a timed change on a waiting voice starts with the sound. - sound-library + node-catalog + soundfx: 419 tests, green. - e2e 34 on melektaus-atx: ALL PASSED, 31 checks. A 2 s ramp from depth 7 to -5 read speed 0.979, 0.926 and 0.778 half a second on. - Counter-check: folding a timed change on a waiting voice into its start values fails the new test. - The gate did not run. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
ed2b58b3c0 |
Load the sounds of a project ahead of their first play
A sound was fetched and decoded on its first Play sound. That play started late, its beginning was not heard, and a graph that tuned the voice in the meantime had moved on by the time it started. Images and fonts were already loaded when a project opens. Sounds now are too. - projectFiles lists the sounds of a project, and preloadProjectFiles hands them to the new soundLibrary.preload. That covers the editor and the published game. - Preloading opens no audio device. It decodes with an OfflineAudioContext at 48 kHz. The real AudioContext is still made on the first play, and a source resamples where the device runs at another rate. - A run started right after opening a project can still find a sound loading. It then starts late as before. Tests: - sound-library: a preload makes only the offline context, and a play after it starts at once, opening the device only then. - sound-library + project-files: 18 tests, green. - e2e 34 on melektaus-atx: ALL PASSED, 30 checks. - Counter-check: decoding with the real AudioContext again fails the new test. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
c7e9769f90 |
Change the sound FX parameters of a sound while it plays #315
Until now a value handed to a sound FX counted from the start of the voice only. An engine whose pitch follows the speed, or a loop that gets quieter, could not be built. Every play of Play sound now gives out the number of its voice, and a new node lets the knobs of that one voice glide to new values while it plays. - Play sound has a new output voice: the number of this one play, or 0 where nothing was played. Kept in a Number variable it reaches the voice from another event. Two plays of one sound are two voices. - New action node Set sound FX parameters (exec, voice, fx params). It reaches exactly the voice it is given, a detached one too. A voice that has ended, or voice 0, is not reached. - AudioSink.play returns the voice number and takes the plan and the values (FxVoice) instead of a finished chain. AudioSink.tune takes a voice number and new values. A name the voice's FX does not have is left out, one not given keeps its value. - SoundLibrary hands out the number at once, before the bytes are there, and never twice. A voice still waiting for its bytes starts with the values handed in meanwhile. - Every knob of a voice moves with setTargetAtTime (30 ms time constant). A knob whose value did not change is not sent again, so a tune every frame costs next to nothing. - A pitch on a playing voice glides by playbackRate. A new WSOLA copy in the middle of a sound is not worked out, so the voice gets shorter or longer while it is moved. On a loop that is not heard. - fxChainOf now gives a chain whose shape depends on the plan alone. A stage fed by a parameter stays at 0 dB, and a Mix fed by a parameter keeps both ways. A typed-in 0 or Mix at 0/1 is still left out by the compiler. - Every way of a voice has a source of its own, also at the same pitch, so each can glide on its own. The pitched copy is shared. - The tab keeps the numbers of what it plays, so moving a parameter field is heard at once. - Test hook __gm.sound.levels(file) reads the knobs of every voice of a file, oldest first, and stats() counts the voices tuned. Tests: - soundfx.test.ts: 29 tests (+2 for the node in a running world: the voice of the first of two plays of one sound, and a play that played nothing) - sound-library.test.ts: 15 tests (+3: glide, reach by number, pending voice) - node-catalog.test.ts: 370 tests (+2, Play sound checks its voice) - sounds + keep-on-scene-load: 31 tests, unchanged - Counter-test: a tune that reaches every voice of the same file makes the reach test fail. - e2e 34 (melektaus-atx): ALL PASSED - 30 checks (+1). In Chrome two looping voices of one sound start at depth 7. The one whose number sits in a variable glides to 0.7938 (2^(-4/12) = 0.7937), the other stays at 1. Both gains stay at 0.5012. - The gate ran before voices got a number (CHECK EXIT 0, 241 files, 3658 tests), not after. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
072537b2e8 |
Blend two ways of a sound FX with a Mix node #316
A sound FX was a single line from Input to Output, so two versions of one sound could not be heard together. The new Mix node joins two ways again, and each way can carry its own effects, its own Pitch included. - Mix has the audio inputs a and b and a number input mix (0 to 1, default 0.5). At 0 only a is heard, at 1 only b. Like every number it can come from a parameter of the Input and is held to its range. - The compiled plan and the chain a voice gets are now a tree. A source carries a pitch and stages, a mix carries a, b and the stages after it. A Mix that ends at exactly 0 or 1 collapses to the one way heard, so the silent way is never worked out. - A Pitch is still worked out on the sound as it starts (#312). Behind a Mix it goes to both of its sounds. An effect before a Pitch therefore acts as if it came after it, the node now says so. - The player starts one source per pitch, all at once and all of the same length, and two ways at the same pitch share one. A mix becomes two gain nodes with 1 - mix and mix into a sum. A voice leaves the count when all of its sources have ended. - stats().last reports pitches (one per source) instead of pitch. Tests: - soundfx.test.ts: 27 tests (+5), sound-library.test.ts: 12 (+2), node-catalog.test.ts: 368, all green. Typecheck, ESLint, Prettier clean. - Counter-check: a Pitch behind the Mix given to a only, and b mixed at 1 - mix, each fails its new test. - e2e 34 on melektaus-atx: ALL PASSED, 29 checks (+1). The new check plays the pitched way mixed with the dry Input in the browser: two sources at 5 and 0 semitones, as long as the sound itself. - The full gate did not run before this commit. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
790e483198 |
Pitch a sound FX without changing its length #312 #313 #314
The Pitch node changed the playback speed, so a sound pitched up was also shorter and one pitched down longer. A pitch now keeps the length, takes fractions and reaches further. Gain and the EQ bands reach further down, and the EQ is renamed so a parametric one can come later without a mix-up. - Pitch shifting is the pure function pitchShift in core/soundfx/pitchShift.ts. It reads the sound faster or slower and stretches it back by WSOLA: 40 ms frames under a Hann window, each moved within 12.5 ms to where it continues the frame before it. A looping sound reads on across its end. - SoundLibrary plays a pitched copy of the decoded buffer instead of setting playbackRate. Copies are kept per file, pitch and loop, the last 32 of them. - stats().last tells the duration of what was played, so a check can see that the length is kept. - Pitch: -64 to +64 semitones, scrubbed in steps of 0.1 (#312). - Gain: -120 to +24 dB (#313). - EQ is now labelled "5-Band-EQ", its bands go -60 to +24 dB (#314). The node type stays SoundFxEq, so saved graphs load unchanged. Findings: - A plain granular shifter was tried first and missed the pitch: at +0.5 semitones it measured 438 Hz instead of 452.9 Hz, because every grain restarts at the phase of the original. WSOLA measures 452 Hz. - Far ends: +64 semitones from 440 Hz lands above half the sample rate and folds back, -64 is below what a 40 ms frame can follow. Nobody has listened to either. - Speed: 3 s of a 48 kHz channel take 47 ms at +3 and +64, 171 ms at -64 (measured in Node). Tests: - pitch-shift (new): 7 tests. A 440 Hz sine at +12, -12, +7, +0.5 and -30.5 semitones is within 2 % of the target and keeps its length. Level stays at most 1, a loop does not fade at its seam. - sound-library: 10 tests (+1), soundfx 22, node-catalog 368. - Counter-check: with the pitched copy switched off the new sound-library test fails. - e2e 34 (melektaus-atx): ALL PASSED - 28 checks. Its octave-down voice is now also checked for the length of the sound. - Typecheck, ESLint and Prettier clean. The gate did not run. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
adminandClaude Opus 5.5
|
de14f0f9e2 |
Play a sound through a graph of gain, pitch and EQ effects #311
A sound could only be played as it was uploaded. A Sound FX asset is
a node graph like a shader: Input, effects, Output. Play sound has a
new input fx that sends the voice through it.
- New asset type soundfx ("Sound FX"), not placeable, with a graph
and its own tab. The heading in the asset panel is "Sound effects",
so it does not take the click meant for the + menu entry.
- Nodes in core/soundfx/nodes.ts, category sound, port type audio
that only fits audio: Input, Output, Gain (dB), Pitch (semitones),
EQ (80 Hz and 12 kHz shelves, peaks at 250 Hz, 1 kHz and 4 kHz).
- compileSoundFx walks back from the Output with the shared graph
emitter. An input takes one edge, so the result is one line: a
pitch and a list of gain and filter stages. A node without input
or a graph without Output reports a problem on the node.
- Pitch is the playback rate of the source. It applies to the whole
sound and also changes its length, the node description says so.
- The BehaviorSystem resolves fx and hands the line to the AudioSink.
An fx that is missing or does not compile plays the sound dry.
- SoundLibrary wires source, stages, volume, destination in that
order. stats().last now also names pitch and stage count.
- The tab header picks one of the project's sounds and plays it
through the graph as it stands. Stop there ends only its own voice.
Tests:
- soundfx.test.ts (new, 11), sound-library (9, +2 for the wiring),
node-catalog (+1 for fx), graph and asset-groups adjusted
- e2e 34: ALL PASSED - 26 checks (+5: + menu opens the tab, palette,
2 effects, tab Play and game Play through the FX at -12 st)
- Gate on melektaus-atx: typecheck, lint, prettier clean, 239 of 240
test files green. The red one was sound-library (its fake
AudioContext had no playbackRate), fixed afterwards, gate not
re-run in full.
- Counter-check: without the audio rule in portsCompatible the port
test fails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
admin
|
1e9d21cd5d |
A game that is more than one scene needs a way from one to the next, so the graph can load one (#39)
game-maker / suite (push) Failing after 11m6s
Load scene sets a flag on the World like Stop and Restart do; the swap happens after the tick, and the last wish of a tick wins. Exactly one thing crosses over: the SharedState, so global variables and global machines keep what they know while everything local is built anew. Loading the scene that is already running rebuilds it, which is what "reset the level" means when the score must survive. A run starts either from the start scene (a new project field, set from the asset panel and marked with an arrow in the list) or from whatever scene is open. The toggle next to Play says which question is asked and remembers the answer in localStorage: it is a way of working, not part of the game. Sounds no longer end because a scene ends. A voice belongs to the behavior that started it and stops right after On destroy, so the scene swap is no longer a special case at all. For the enemy that plays its explosion and destroys itself in the same breath there is now a detached input on Play sound. The scene selector stays usable while a run plays, because a double click in the asset panel already did the same thing unguarded. It moves what the editor looks at, not what the run plays. |
||
|
admin
|
5671a38044 |
A game that stays silent is only half a game, so a sound is an asset like any other (#45)
game-maker / suite (push) Successful in 12m1s
Sounds follow the road fonts already paved: the file goes to the server content-addressed under its sha256, the project keeps a named asset that points at it, and nothing is ever re-encoded. The format is read from the file's own signature, not from what the browser claims - the same .m4a arrives as audio/mp4, audio/x-m4a or as nothing at all. Playing is two nodes. "Play sound" takes the asset and may raise or lower its volume; every call is its own voice, so two jumps in a row are two jumps heard. Its loop input is a choice of three - from the asset, on, off - because a checkbox has to spend one of its two positions on "not set" and would lose a case: an asset that loops could then never be played once. Below the graph core stays browser-free: NodeContext hands out an asset id, BehaviorSystem resolves it against the project and folds the asset's own volume and loop into the call, and only viewport/SoundLibrary knows about Web Audio. A run without a sink plays nothing instead of throwing. The CI browser is muted, so the ear is no witness: the debug hook counts started and stopped voices, and step 34 reads that counter. |