Nodejs sdk

.github/workflows/nodejs-perf.yml

@@ -6,15 +6,13 @@ on:
     branches: [ main ]
     paths:
       - 'binding/nodejs/perf.test.js'
-      - 'lib/common/**'
       - 'lib/node/**'
       - '.github/workflows/nodejs-perf.yml'
 
   pull_request:
     branches: [ main, 'v[0-9]+.[0-9]+' ]
     paths:
       - 'binding/nodejs/perf.test.js'
-      - 'lib/common/**'
       - 'lib/node/**'
       - '.github/workflows/nodejs-perf.yml'
 
@@ -33,15 +31,12 @@ jobs:
         include:
           - os: ubuntu-latest
             num_test_iterations: 50
-            init_performance_threshold_sec: 0.001
             proc_performance_threshold_sec: 0.005
           - os: windows-latest
             num_test_iterations: 50
-            init_performance_threshold_sec: 0.5
             proc_performance_threshold_sec: 0.005
           - os: macos-latest
             num_test_iterations: 50
-            init_performance_threshold_sec: 0.005
             proc_performance_threshold_sec: 0.005
 
     steps:
@@ -56,7 +51,7 @@ jobs:
         run: yarn install
 
       - name: Test
-        run: yarn test perf.test.ts --access_key=${{secrets.PV_VALID_ACCESS_KEY}} --num_test_iterations=${{matrix.num_test_iterations}} --init_performance_threshold_sec=${{matrix.init_performance_threshold_sec}} --proc_performance_threshold_sec=${{matrix.proc_performance_threshold_sec}}
+        run: yarn test perf.test.ts --access_key=${{secrets.PV_VALID_ACCESS_KEY}} --num_test_iterations=${{matrix.num_test_iterations}} --proc_performance_threshold_sec=${{matrix.proc_performance_threshold_sec}}
 
   perf-self-hosted:
     runs-on: ${{ matrix.machine }}
@@ -97,7 +92,7 @@ jobs:
         run: bash machine-state.sh
 
       - name: Test
-        run: yarn test perf.test.ts --access_key=${{secrets.PV_VALID_ACCESS_KEY}} --num_test_iterations=${{matrix.num_test_iterations}} --init_performance_threshold_sec=0.01 --proc_performance_threshold_sec=0.05
+        run: yarn test perf.test.ts --access_key=${{secrets.PV_VALID_ACCESS_KEY}} --num_test_iterations=${{matrix.num_test_iterations}} --proc_performance_threshold_sec=0.05
 
       - name: Machine state after
         working-directory: res/scripts

README.md

@@ -347,10 +347,10 @@ Create instances of the Cobra class:
 const { Cobra } = require("@picovoice/cobra-node");
 
 const accessKey = "${ACCESS_KEY}"; // Obtained from the Picovoice Console (https://console.picovoice.ai/)
-const handle = new Cobra(accessKey);
+const cobra = new Cobra(accessKey);
 ```
 
-When instantiated, `handle` can process audio via its `.process` method.
+When instantiated, `cobra` can process audio via its `.process` method.
 
 ```javascript
 function getNextAudioFrame() {
@@ -360,15 +360,15 @@ function getNextAudioFrame() {
 
 while (true) {
   const audioFrame = getNextAudioFrame();
-  const voiceProbability = handle.process(audioFrame);
+  const voiceProbability = cobra.process(audioFrame);
   console.log(voiceProbability);
 }
 ```
 
 When done be sure to release resources using `release()`:
 
 ```javascript
-handle.release();
+cobra.release();
 ```
 
 ### Rust

binding/nodejs/README.md

@@ -35,34 +35,36 @@ Signup or Login to [Picovoice Console](https://console.picovoice.ai/) to get you
 
 Create an instance of the engine and detect voice activity in the audio:
 
-Replace `${AccessKey}` with your AccessKey obtained from [Picovoice Console](https://console.picovoice.ai/). `handle` is
+Replace `${AccessKey}` with your AccessKey obtained from [Picovoice Console](https://console.picovoice.ai/). `cobra` is
 an instance of Cobra.
 
-When initialized, the valid sample rate is given by `handle.sampleRate`. Expected frame length (number of audio samples
-in an input array) is `handle.frameLength`. The engine accepts 16-bit linearly-encoded PCM and operates on
-single-channel audio.
-
 ```javascript
 const { Cobra } = require("@picovoice/cobra-node");
 
 const accessKey = "${ACCESS_KEY}"; // Obtained from the Picovoice Console (https://console.picovoice.ai/)
-const handle = new Cobra(accessKey);
+const cobra = new Cobra(accessKey);
+```
 
+Once initialized, the valid sample rate is given by `cobra.sampleRate`. The expected frame length (number of audio samples
+in an input array) is `cobra.frameLength`. The engine accepts 16-bit linearly-encoded PCM and operates on
+single-channel audio.
+
+```javascript
 function getNextAudioFrame() {
   // ...
   return audioFrame;
 }
 
 while (true) {
   const audioFrame = getNextAudioFrame();
-  const voiceProbability = handle.process(audioFrame);
+  const voiceProbability = cobra.process(audioFrame);
   console.log(voiceProbability);
 }
 ```
 
 Finally, when done be sure to explicitly release the resources using
-`handle.release()`.
+`cobra.release()`.
 
 ## Demos
 
-[Cobra Node.js demo package](../../demo/nodejs) provides command-line utilities for processing audio using cobra.
+The [Cobra Node.js demo package](../../demo/nodejs) provides command-line utilities for processing audio using Cobra.

binding/nodejs/package.json

@@ -9,7 +9,6 @@
     "build:all": "tsc",
     "build:types": "tsc --declaration --declarationMap --emitDeclarationOnly --outDir ./dist/types",
     "prepack": "npm run build",
-    "prepare": "node copy.js",
     "test": "jest --no-cache",
     "lint": "eslint . --ext .js,.ts"
   },

binding/nodejs/src/cobra.ts

@@ -119,7 +119,7 @@ export default class Cobra {
    *
    * @param {Int16Array} pcm Audio data. The audio needs to have a sample rate equal to `Cobra.sampleRate` and be 16-bit linearly-encoded.
    * The specific array length can be attained by calling `Cobra.frameLength`. This function operates on single-channel audio.
-   * @returns {number} Voice probability result.
+   * @returns {number} Probability of voice activity. It is a floating-point number within [0, 1].
    */
   process(pcm: Int16Array): number {
     assert(pcm instanceof Int16Array);
@@ -142,19 +142,19 @@ export default class Cobra {
       );
     }
 
-    let VoiceProbabilityAndStatus: VoiceProbabilityAndStatus | null = null;
+    let voiceProbabilityAndStatus: VoiceProbabilityAndStatus | null = null;
     try {
-      VoiceProbabilityAndStatus = this._pvCobra.process(this._handle, pcm);
+      voiceProbabilityAndStatus = this._pvCobra.process(this._handle, pcm);
     } catch (err: any) {
       pvStatusToException(PvStatus[err.code as keyof typeof PvStatus], err);
     }
 
-    const status = VoiceProbabilityAndStatus!.status;
+    const status = voiceProbabilityAndStatus!.status;
     if (status !== PvStatus.SUCCESS) {
       this.handlePvStatus(status, 'Cobra failed to process the audio frame');
     }
 
-    return VoiceProbabilityAndStatus!.is_voiced;
+    return voiceProbabilityAndStatus!.is_voiced;
   }
 
   /**

binding/nodejs/test/perf.test.ts

@@ -28,39 +28,13 @@ const NUM_TEST_ITERATIONS = Number(
     .filter(x => x.startsWith('--num_test_iterations='))[0]
     ?.split('--num_test_iterations=')[1] ?? 0
 );
-const INIT_PERFORMANCE_THRESHOLD_SEC = Number(
-  process.argv
-    .filter(x => x.startsWith('--init_performance_threshold_sec='))[0]
-    ?.split('--init_performance_threshold_sec=')[1] ?? 0
-);
 const PROC_PERFORMANCE_THRESHOLD_SEC = Number(
   process.argv
     .filter(x => x.startsWith('--proc_performance_threshold_sec='))[0]
     ?.split('--proc_performance_threshold_sec=')[1] ?? 0
 );
 
 describe('Performance', () => {
-  test('init performance', () => {
-    let perfResults = [];
-    for (let i = 0; i < NUM_TEST_ITERATIONS; i++) {
-      const before = performance.now();
-      let cobraEngine = new Cobra(ACCESS_KEY);
-      let initTime = performance.now() - before;
-
-      cobraEngine.release();
-
-      if (i > 0) {
-        perfResults.push(initTime);
-      }
-    }
-    let avgPerfMs =
-      perfResults.reduce((acc, a) => acc + a, 0) / NUM_TEST_ITERATIONS;
-    let avgPerfSec = Number((avgPerfMs / 1000).toFixed(3));
-    // eslint-disable-next-line no-console
-    console.log('Average init performance: ' + avgPerfSec);
-    expect(avgPerfSec).toBeLessThanOrEqual(INIT_PERFORMANCE_THRESHOLD_SEC);
-  });
-
   test('proc performance', () => {
     let cobraEngine = new Cobra(ACCESS_KEY);
 

demo/nodejs/README.md

@@ -44,6 +44,14 @@ cobra-file-demo --access_key ${ACCESS_KEY} --input_audio_file_path ${AUDIO_PATH}
 Replace `${ACCESS_KEY}` with yours obtained from Picovoice Console and `${AUDIO_PATH}` with a path to an audio file you
 wish to use for voice activity detection.
 
+The threshold of the engine can be tuned using the `threshold` input argument:
+
+```console
+cobra-file-demo --access_key ${ACCESS_KEY} --input_audio_file_path ${AUDIO_PATH} --threshold ${DETECTION_THRESHOLD}
+```
+
+Threshold is a floating point number within `[0, 1]`. A higher threshold reduces the miss rate at the cost of increased false alarm rate.
+
 ### Microphone Demo
 
 You need a working microphone connected to your machine for this demo. Run the following in the terminal:

demo/nodejs/file.js

@@ -31,6 +31,10 @@ program
   .option(
     "-l, --library_file_path <string>",
     "absolute path to cobra dynamic library"
+  )
+  .option(
+    "-t --threshold <string>",
+    "Threshold for the probability of voice activity"
   );
 
 if (process.argv.length < 2) {
@@ -42,6 +46,7 @@ function fileDemo() {
   let audioPath = program["input_audio_file_path"];
   let accessKey = program["access_key"];
   let libraryFilePath = program["library_file_path"];
+  let threshold = program["threshold"] ?? 0.8;
 
   let engineInstance = new Cobra(accessKey, {
     libraryPath: libraryFilePath,
@@ -70,13 +75,20 @@ function fileDemo() {
 
   let frames = getInt16Frames(inputWaveFile, engineInstance.frameLength);
 
-  let voiceProbabilities = [];
-  for (let frame of frames) {
-    const voiceProbability = engineInstance.process(frame);
-    voiceProbabilities.push(voiceProbability);
+  const printedNums = new Set();
+  for (let i = 0; i < frames.length; i++) {
+    const result = engineInstance.process(frames[i]);
+    const timestamp = (
+      (i * engineInstance.frameLength) /
+      engineInstance.sampleRate
+    ).toFixed(1);
+
+    if (result >= threshold && !printedNums.has(timestamp)) {
+      console.log(`Detected voice activity at ${timestamp} sec`);
+      printedNums.add(timestamp);
+    }
   }
 
-  console.log(voiceProbabilities);
   engineInstance.release();
 }
 

demo/nodejs/mic.js

@@ -90,7 +90,16 @@ async function micDemo() {
     const pcm = await recorder.read();
     try {
       const voiceProbability = engineInstance.process(pcm);
-      console.log(voiceProbability);
+      const percentage = voiceProbability * 100;
+      const barLength = Math.floor((percentage / 10) * 3);
+      const emptyLength = 30 - barLength;
+      const spacer = ` `.repeat(3 - percentage.toFixed(0).length);
+
+      process.stdout.write(
+        `\r[${spacer}${percentage.toFixed(0)}]|${"█".repeat(
+          barLength
+        )}${" ".repeat(emptyLength)}|`
+      );
     } catch (err) {
       if (err instanceof CobraActivationLimitReachedError) {
         console.error(
@@ -106,6 +115,7 @@ async function micDemo() {
   recorder.stop();
   recorder.release();
   engineInstance.release();
+  process.stdout.write("\n");
   process.exit();
 }