@newgameplusinc/odyssey-audio-video-sdk-dev 1.0.24 → 1.0.26

package/README.md

@@ -83,6 +83,8 @@ sdk.setListenerFromLSD(listenerPos, cameraPos, lookAtPos);
 
 #### Noise-Cancellation Stack (What’s Included)
 1. **Adaptive denoiser worklet** – learns each participant’s noise floor in real time, then applies a multi-band downward expander plus dynamic low/high-pass shaping.
+	- `speechBoost` lifts the low/mid band only when speech confidence is high, keeping consonants bright without reintroducing floor noise.
+	- `highBandGate` + `highBandAttack/Release` clamp constant fan hiss in the 4–12 kHz band whenever speechPresence is low, so background whoosh never leaks through live mics.
 2. **Optional voice enhancement** – autocorrelation-derived confidence (inspired by the tuner article) can raise the reduction floor when speech is present to keep vocals bright.
 3. **Silence gate** – if energy stays below `silenceFloor` for a configurable hold window, the track ramps to true silence, then wakes instantly once voice energy returns.
 4. **Classic filters** – fixed high-pass/low-pass filters shave off rumble and hiss before signals reach the HRTF panner.
@@ -91,16 +93,20 @@ These layers run entirely in Web Audio, so you can ship “AirPods-style” back
 	```ts
 	const sdk = new OdysseySpatialComms(serverUrl, {
 		denoiser: {
-			threshold: 0.0085,
-			maxReduction: 0.9,
-			hissCut: 0.55,
-			holdMs: 230,
+			threshold: 0.008,
+			maxReduction: 0.88,
+			hissCut: 0.52,
+			holdMs: 260,
 			voiceBoost: 0.65,
 			voiceSensitivity: 0.33,
 			voiceEnhancement: true,
-			silenceFloor: 0.0008,
-			silenceHoldMs: 480,
-			silenceReleaseMs: 180,
+			silenceFloor: 0.00075,
+			silenceHoldMs: 520,
+			silenceReleaseMs: 160,
+			speechBoost: 0.35,
+			highBandGate: 0.7,
+			highBandAttack: 0.25,
+			highBandRelease: 0.12,
 		},
 	});
 	```

package/dist/SpatialAudioManager.d.ts

@@ -23,6 +23,10 @@ type DenoiserOptions = {
     silenceFloor?: number;
     silenceHoldMs?: number;
     silenceReleaseMs?: number;
+    speechBoost?: number;
+    highBandGate?: number;
+    highBandAttack?: number;
+    highBandRelease?: number;
 };
 export type SpatialAudioOptions = {
     distance?: SpatialAudioDistanceConfig;

package/dist/SpatialAudioManager.js

@@ -94,6 +94,10 @@ class SpatialAudioManager extends EventManager_1.EventManager {
                         silenceFloor: this.options.denoiser?.silenceFloor,
                         silenceHoldMs: this.options.denoiser?.silenceHoldMs,
                         silenceReleaseMs: this.options.denoiser?.silenceReleaseMs,
+                        speechBoost: this.options.denoiser?.speechBoost,
+                        highBandGate: this.options.denoiser?.highBandGate,
+                        highBandAttack: this.options.denoiser?.highBandAttack,
+                        highBandRelease: this.options.denoiser?.highBandRelease,
                     },
                 });
             }
@@ -603,8 +607,13 @@ class SpatialAudioManager extends EventManager_1.EventManager {
     this.voiceBoost = this._sanitize(cfg.voiceBoost, 0, 1, 0.6);
     this.voiceSensitivity = this._sanitize(cfg.voiceSensitivity, 0.05, 0.9, 0.35);
     this.voiceEnhancement = cfg.voiceEnhancement === true;
+    this.speechBoost = this._sanitize(cfg.speechBoost, 0, 1.5, 0.35);
+    this.highBandGate = this._sanitize(cfg.highBandGate, 0, 1, 0.7);
+    this.highBandAttack = this._sanitize(cfg.highBandAttack, 0.01, 1, 0.25);
+    this.highBandRelease = this._sanitize(cfg.highBandRelease, 0.01, 1, 0.12);
     this.silenceFloor = this._sanitize(cfg.silenceFloor, 0.0002, 0.02, 0.00085);
     this.gateGraceSamples = Math.round(sampleRate * 0.45);
+    this.postSpeechHoldSamples = Math.round(sampleRate * 0.35);
     this.silenceHoldSamples = Math.max(
       8,
       Math.round(
@@ -647,6 +656,9 @@ class SpatialAudioManager extends EventManager_1.EventManager {
         isSilenced: false,
         muteGain: 1,
         graceSamplesRemaining: this.gateGraceSamples,
+        postSpeechHold: 0,
+        highBandEnv: this.silenceFloor,
+        broadbandEnv: this.silenceFloor,
       };
     }
     return this.channelState[index];
@@ -743,6 +755,13 @@ class SpatialAudioManager extends EventManager_1.EventManager {
 
         state.envelope += (magnitude - state.envelope) * this.attack;
 
+        if (speechPresence > 0.12 || state.envelope > this.threshold * 1.1) {
+          state.graceSamplesRemaining = this.gateGraceSamples;
+          state.postSpeechHold = this.postSpeechHoldSamples;
+        } else if (state.postSpeechHold > 0) {
+          state.postSpeechHold--;
+        }
+
         if (state.envelope < this.threshold) {
           state.noise += (state.envelope - state.noise) * this.learnRate;
           state.quietSamples++;
@@ -750,14 +769,14 @@ class SpatialAudioManager extends EventManager_1.EventManager {
           state.quietSamples = 0;
         }
 
-        if (state.graceSamplesRemaining > 0) {
+        if (state.graceSamplesRemaining > 0 || state.postSpeechHold > 0) {
           state.graceSamplesRemaining--;
           state.isSilenced = false;
           state.silenceSamples = 0;
           state.silenceReleaseCounter = 0;
         } else {
           const belowFloor = state.envelope < this.silenceFloor;
-          if (belowFloor && speechPresence < 0.25) {
+          if (belowFloor && speechPresence < 0.15) {
             state.silenceSamples++;
           } else {
             state.silenceSamples = Math.max(0, state.silenceSamples - 3);
@@ -769,12 +788,14 @@ class SpatialAudioManager extends EventManager_1.EventManager {
           }
 
           if (state.isSilenced) {
-            const wakeFromEnergy = state.envelope > this.silenceFloor * 1.25;
-            const wakeFromVoice = speechPresence > 0.2;
+            const wakeFromEnergy = state.envelope > this.silenceFloor * 1.2;
+            const wakeFromVoice = speechPresence > 0.15;
             if (wakeFromEnergy || wakeFromVoice) {
               state.isSilenced = false;
               state.silenceSamples = 0;
               state.silenceReleaseCounter = 0;
+              state.postSpeechHold = this.postSpeechHoldSamples;
+              state.graceSamplesRemaining = this.gateGraceSamples;
             } else {
               state.silenceReleaseCounter++;
               if (state.silenceReleaseCounter > this.silenceReleaseSamples) {
@@ -809,16 +830,31 @@ class SpatialAudioManager extends EventManager_1.EventManager {
         let processed = sample * state.gain;
 
         state.lpState = this.hfAlpha * state.lpState + (1 - this.hfAlpha) * processed;
-        const highComponent = processed - state.lpState;
+        const lowComponent = state.lpState;
+        const highComponent = processed - lowComponent;
+
+        // Track broadband/high-band envelopes for multi-band gating
+        state.broadbandEnv += (Math.abs(processed) - state.broadbandEnv) * 0.12;
+        const highEnvDelta = Math.abs(highComponent) - state.highBandEnv;
+        const highEnvCoef = highEnvDelta > 0 ? this.highBandAttack : this.highBandRelease;
+        state.highBandEnv += highEnvDelta * highEnvCoef;
+
         const hissRatio = Math.min(
           1,
-          Math.abs(highComponent) / (Math.abs(state.lpState) + 1e-5)
+          Math.abs(highComponent) / (Math.abs(lowComponent) + 1e-5)
         );
         const hissGain = 1 - hissRatio * (this.hissCut * (1 - 0.4 * speechPresence));
-        processed = state.lpState + highComponent * hissGain;
 
-        const muteTarget = state.isSilenced ? 0 : 1;
-        state.muteGain += (muteTarget - state.muteGain) * 0.18;
+        const highEnvRatio = state.highBandEnv / (state.broadbandEnv + 1e-5);
+        const gateAmount = this.highBandGate * Math.max(0, highEnvRatio - speechPresence * 0.5);
+        const gatedHigh = highComponent * hissGain * (1 - gateAmount);
+
+        const speechLift = 1 + this.speechBoost * speechPresence;
+        processed = lowComponent * speechLift + gatedHigh;
+
+        const muteTarget = state.isSilenced ? 0.05 : 1;
+        const smoothing = state.isSilenced ? 0.12 : 0.42;
+        state.muteGain += (muteTarget - state.muteGain) * smoothing;
         processed *= state.muteGain;
 
         outChannel[i] = processed;
@@ -858,21 +894,25 @@ registerProcessor('odyssey-denoise', OdysseyDenoiseProcessor);
         };
         const denoiserDefaults = {
             enabled: true,
-            threshold: 0.0085,
-            noiseFloor: 0.0023,
-            release: 0.24,
-            attack: 0.34,
-            holdMs: 230,
-            maxReduction: 0.9,
-            hissCut: 0.55,
-            expansionRatio: 2.2,
-            learnRate: 0.05,
+            threshold: 0.008,
+            noiseFloor: 0.002,
+            release: 0.22,
+            attack: 0.36,
+            holdMs: 260,
+            maxReduction: 0.88,
+            hissCut: 0.52,
+            expansionRatio: 2.15,
+            learnRate: 0.045,
             voiceBoost: 0.65,
             voiceSensitivity: 0.33,
             voiceEnhancement: false,
-            silenceFloor: 0.0008,
-            silenceHoldMs: 480,
-            silenceReleaseMs: 180,
+            silenceFloor: 0.00075,
+            silenceHoldMs: 520,
+            silenceReleaseMs: 160,
+            speechBoost: 0.35,
+            highBandGate: 0.7,
+            highBandAttack: 0.25,
+            highBandRelease: 0.12,
         };
         return {
             distance: {
@@ -898,6 +938,10 @@ registerProcessor('odyssey-denoise', OdysseyDenoiseProcessor);
                 silenceFloor: options?.denoiser?.silenceFloor ?? denoiserDefaults.silenceFloor,
                 silenceHoldMs: options?.denoiser?.silenceHoldMs ?? denoiserDefaults.silenceHoldMs,
                 silenceReleaseMs: options?.denoiser?.silenceReleaseMs ?? denoiserDefaults.silenceReleaseMs,
+                speechBoost: options?.denoiser?.speechBoost ?? denoiserDefaults.speechBoost,
+                highBandGate: options?.denoiser?.highBandGate ?? denoiserDefaults.highBandGate,
+                highBandAttack: options?.denoiser?.highBandAttack ?? denoiserDefaults.highBandAttack,
+                highBandRelease: options?.denoiser?.highBandRelease ?? denoiserDefaults.highBandRelease,
             },
         };
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@newgameplusinc/odyssey-audio-video-sdk-dev",
-  "version": "1.0.24",
+  "version": "1.0.26",
   "description": "Odyssey Spatial Audio & Video SDK using MediaSoup for real-time communication",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",