refactor: rmsDB -> value, use max power in buffer

Author: jamsch

android/src/main/java/expo/modules/speechrecognition/ExpoSpeechService.kt

@@ -466,11 +466,11 @@ class ExpoSpeechService(
         val intervalMs = options.volumeChangeEventOptions?.intervalMillis
 
         if (intervalMs == null) {
-            sendEvent("volumechange", mapOf("rmsdB" to rmsdB))
+            sendEvent("volumechange", mapOf("value" to rmsdB))
         } else {
             val currentTime = System.currentTimeMillis()
             if (currentTime - lastVolumeChangeEventTime >= intervalMs) {
-                sendEvent("volumechange", mapOf("rmsdB" to rmsdB))
+                sendEvent("volumechange", mapOf("value" to rmsdB))
                 lastVolumeChangeEventTime = currentTime
             }
         }

example/components/VolumeMeteringAvatar.tsx

@@ -44,12 +44,13 @@ export function VolumeMeteringAvatar() {
 
   useSpeechRecognitionEvent("volumechange", (event) => {
     // Don't animate anything if the volume is too low
-    if (event.rmsdB <= 1) {
+    if (event.value <= 1) {
       return;
     }
+
     const newScale = interpolate(
-      event.rmsdB,
-      [-2, 10], // The rmsDB range is between -2 and 10
+      event.value,
+      [-2, 10], // The value range is between -2 and 10
       [minScale, maxScale],
       Extrapolation.CLAMP,
     );
@@ -61,7 +62,7 @@ export function VolumeMeteringAvatar() {
         stiffness: 150,
       }),
       withTiming(minScale, {
-        duration: 1000,
+        duration: 500,
         easing: Easing.linear,
       }),
     );
@@ -74,7 +75,7 @@ export function VolumeMeteringAvatar() {
           easing: Easing.out(Easing.quad),
         }),
         withTiming(minScale, {
-          duration: 400,
+          duration: 300,
           easing: Easing.linear,
         }),
       );

example/ios/Podfile.lock

@@ -39,7 +39,7 @@ PODS:
     - ReactCommon/turbomodule/bridging
     - ReactCommon/turbomodule/core
     - Yoga
-  - ExpoSpeechRecognition (0.2.21):
+  - ExpoSpeechRecognition (0.2.22):
     - ExpoModulesCore
   - EXSplashScreen (0.27.5):
     - DoubleConversion
@@ -1466,8 +1466,8 @@ SPEC CHECKSUMS:
   ExpoFileSystem: 80bfe850b1f9922c16905822ecbf97acd711dc51
   ExpoFont: 00756e6c796d8f7ee8d211e29c8b619e75cbf238
   ExpoKeepAwake: 3b8815d9dd1d419ee474df004021c69fdd316d08
-  ExpoModulesCore: a113755f96c40590671f01cfcdce8ebdf0cf5f83
-  ExpoSpeechRecognition: 80eefd4f4bd7541f5fad24744b19a4b36a365414
+  ExpoModulesCore: 260ee156852434da26e782bbb993093186c5aade
+  ExpoSpeechRecognition: 6ce34ee84af8950f0e472ba6f6b0b3def2553b2a
   EXSplashScreen: a7e8d13c476f9937e39d654af4235758b567a1be
   FBLazyVector: ac12dc084d1c8ec4cc4d7b3cf1b0ebda6dab85af
   fmt: 4c2741a687cc09f0634a2e2c72a838b99f1ff120

ios/ExpoSpeechRecognitionModule.swift

@@ -185,8 +185,8 @@ public class ExpoSpeechRecognitionModule: Module {
                 self?.sendEvent("audioend", ["uri": nil])
               }
             },
-            volumeChangeHandler: { [weak self] rmsdB in
-                self?.sendEvent("volumechange", ["rmsdB": rmsdB])
+            volumeChangeHandler: { [weak self] value in
+              self?.sendEvent("volumechange", ["value": value])
             }
           )
         } catch {

ios/ExpoSpeechRecognizer.swift

@@ -607,23 +607,6 @@ actor ExpoSpeechRecognizer: ObservableObject {
     try audioEngine.start()
   }
 
-  //  private static func calculatePower(buffer: AVAudioPCMBuffer) -> Float? {
-  //    guard let channelData = buffer.floatChannelData?[0] else { return nil }
-  //    let frameLength = Int(buffer.frameLength)
-  //
-  //    var sumOfSquares: Float = 0.0
-  //    for i in 0..<frameLength {
-  //      let sample = channelData[i]
-  //      sumOfSquares += sample * sample
-  //    }
-  //
-  //    let meanSquare = sumOfSquares / Float(frameLength)
-  //    let rmsEpsilon: Float = 1e-7  // -160 dB
-  //    let rms = sqrt(meanSquare + rmsEpsilon)
-  //
-  //    return 20.0 * log10(rms)
-  //  }
-
   private static func calculatePower(buffer: AVAudioPCMBuffer) -> Float? {
     // let channelCount = Int(buffer.format.channelCount)
     let length = vDSP_Length(buffer.frameLength)
@@ -659,14 +642,7 @@ actor ExpoSpeechRecognizer: ObservableObject {
     if max < kMinLevel {
       max = kMinLevel
     }
-
-    var rms: Float = 0.0
-    vDSP_rmsqv(data, strideFrames, &rms, length)
-    if rms < kMinLevel {
-      rms = kMinLevel
-    }
-
-    return 20.0 * log10(rms)
+    return 20.0 * log10(max)
   }
 
   /// Downsamples the audio buffer to a file ref

src/ExpoSpeechRecognitionModule.types.ts

@@ -129,11 +129,11 @@ export type ExpoSpeechRecognitionNativeEventMap = {
   soundend: null;
   volumechange: {
     /**
-     * A value between -2 and 10 indicating the volume of the input audio
+     * A float value between -2 and 10 indicating the volume of the input audio
      *
      * Consider anything below 0 to be inaudible
      */
-    rmsdB: number;
+    value: number;
   };
 };