@digipair/skill-web-voice-activity-detection 0.94.0-0 → 0.94.0-3

package/index.cjs.js

@@ -0,0 +1,110 @@
+'use strict';
+
+Object.defineProperty(exports, '__esModule', { value: true });
+
+var engine = require('@digipair/engine');
+
+function _extends() {
+    _extends = Object.assign || function assign(target) {
+        for(var i = 1; i < arguments.length; i++){
+            var source = arguments[i];
+            for(var key in source)if (Object.prototype.hasOwnProperty.call(source, key)) target[key] = source[key];
+        }
+        return target;
+    };
+    return _extends.apply(this, arguments);
+}
+
+// add a script tag to the document
+async function addScript(src) {
+    const script = document.createElement('script');
+    script.src = src;
+    script.async = false;
+    document.body.appendChild(script);
+    return new Promise((resolve)=>{
+        script.onload = resolve;
+    });
+}
+let VADService = class VADService {
+    async listen(params, _pinsSettingsList, context) {
+        const globalInstance = typeof window === 'undefined' ? global : window;
+        const config = globalInstance.__DIGIPAIR_CONFIG__;
+        const { stream = await navigator.mediaDevices.getUserMedia({
+            audio: {
+                channelCount: 1,
+                echoCancellation: true,
+                autoGainControl: true,
+                noiseSuppression: true
+            }
+        }), onFrameProcessed = [], onVADMisfire = [], onSpeechStart = [], onSpeechEnd = [], model = 'legacy', additionalAudioConstraints = {}, baseAssetPath = `${config.BASE_URL}/@ricky0123/vad-web@0.0.22/dist/`, onnxWASMBasePath = `${config.BASE_URL}/onnxruntime-web@1.18.0/dist/`, submitUserSpeechOnPause = false, positiveSpeechThreshold = 0.5, negativeSpeechThreshold = 0.35, preSpeechPadFrames = 1, redemptionFrames = 8, frameSamples = 1536, minSpeechFrames = 3 } = params;
+        if (!document.querySelector(`script[src="${onnxWASMBasePath}ort.js"]`)) {
+            await addScript(`${onnxWASMBasePath}ort.js`);
+        }
+        if (!document.querySelector(`script[src="${baseAssetPath}bundle.min.js"]`)) {
+            await addScript(`${baseAssetPath}bundle.min.js`);
+        }
+        const { MicVAD, utils } = window['vad'];
+        const vad = await MicVAD.new({
+            stream,
+            onFrameProcessed: (probabilities, frame)=>{
+                engine.executePinsList(onFrameProcessed, _extends({}, context, {
+                    probabilities,
+                    frame,
+                    vad
+                }), `${context.__PATH__}.onFrameProcessed`);
+            },
+            onVADMisfire: ()=>{
+                engine.executePinsList(onVADMisfire, _extends({}, context, {
+                    vad
+                }), `${context.__PATH__}.onVADMisfire`);
+            },
+            onSpeechStart: ()=>{
+                engine.executePinsList(onSpeechStart, _extends({}, context, {
+                    vad
+                }), `${context.__PATH__}.onSpeechStart`);
+            },
+            onSpeechEnd: (buffer)=>{
+                const wavBuffer = utils.encodeWAV(buffer);
+                const base64 = utils.arrayBufferToBase64(wavBuffer);
+                const audio = `data:audio/wav;base64,${base64}`;
+                engine.executePinsList(onSpeechEnd, _extends({}, context, {
+                    audio,
+                    vad
+                }), `${context.__PATH__}.onSpeechEnd`);
+            },
+            additionalAudioConstraints,
+            positiveSpeechThreshold,
+            negativeSpeechThreshold,
+            redemptionFrames,
+            frameSamples,
+            preSpeechPadFrames,
+            minSpeechFrames,
+            model,
+            baseAssetPath,
+            onnxWASMBasePath,
+            submitUserSpeechOnPause
+        });
+        return vad;
+    }
+    async start(params, _pinsSettingsList, context) {
+        const { vad = context.vad } = params;
+        return vad.start();
+    }
+    async pause(params, _pinsSettingsList, context) {
+        const { vad = context.vad } = params;
+        return vad.pause();
+    }
+    async destroy(params, _pinsSettingsList, context) {
+        const { vad = context.vad } = params;
+        return vad.destroy();
+    }
+};
+const listen = (params, pinsSettingsList, context)=>new VADService().listen(params, pinsSettingsList, context);
+const start = (params, pinsSettingsList, context)=>new VADService().start(params, pinsSettingsList, context);
+const pause = (params, pinsSettingsList, context)=>new VADService().pause(params, pinsSettingsList, context);
+const destroy = (params, pinsSettingsList, context)=>new VADService().destroy(params, pinsSettingsList, context);
+
+exports.destroy = destroy;
+exports.listen = listen;
+exports.pause = pause;
+exports.start = start;

package/index.d.ts

@@ -0,0 +1 @@
+export * from './libs/skill-web-voice-activity-detection/src/index';