refac: voice recording web api

src/lib/components/chat/MessageInput.svelte

@@ -452,6 +452,10 @@
 
 							await tick();
 							document.getElementById('chat-textarea')?.focus();
+
+							if ($settings?.speechAutoSend ?? false) {
+								submitPrompt(prompt, user);
+							}
 						}}
 					/>
 				{:else}

src/lib/components/chat/MessageInput/VoiceRecording.svelte

@@ -18,6 +18,8 @@
 	let durationSeconds = 0;
 	let durationCounter = null;
 
+	let transcription = '';
+
 	const startDurationCounter = () => {
 		durationCounter = setInterval(() => {
 			durationSeconds++;
@@ -104,15 +106,15 @@
 
 					visualizerData = visualizerData;
 
-					if (domainData.some((value) => value > 0)) {
+					// if (domainData.some((value) => value > 0)) {
-						lastSoundTime = Date.now();
+					// 	lastSoundTime = Date.now();
-					}
+					// }
 
-					if (recording && Date.now() - lastSoundTime > 3000) {
+					// if (recording && Date.now() - lastSoundTime > 3000) {
-						if ($settings?.speechAutoSend ?? false) {
+					// 	if ($settings?.speechAutoSend ?? false) {
-							confirmRecording();
+					// 		confirmRecording();
-						}
+					// 	}
-					}
+					// }
 				}
 
 				window.requestAnimationFrame(processFrame);
@@ -165,7 +167,9 @@
 		mediaRecorder.ondataavailable = (event) => audioChunks.push(event.data);
 		mediaRecorder.onstop = async () => {
 			console.log('Recording stopped');
-
+			if (($settings?.audio?.STTEngine ?? '') === 'web') {
+				audioChunks = [];
+			} else {
 				if (confirmed) {
 					const audioBlob = new Blob(audioChunks, { type: 'audio/wav' });
 
@@ -174,11 +178,70 @@
 					confirmed = false;
 					loading = false;
 				}
-
 				audioChunks = [];
 				recording = false;
+			}
 		};
 		mediaRecorder.start();
+
+		if (($settings?.audio?.STTEngine ?? '') === 'web') {
+			if ('SpeechRecognition' in window || 'webkitSpeechRecognition' in window) {
+				// Create a SpeechRecognition object
+				speechRecognition = new (window.SpeechRecognition || window.webkitSpeechRecognition)();
+
+				// Set continuous to true for continuous recognition
+				speechRecognition.continuous = true;
+
+				// Set the timeout for turning off the recognition after inactivity (in milliseconds)
+				const inactivityTimeout = 3000; // 3 seconds
+
+				let timeoutId;
+				// Start recognition
+				speechRecognition.start();
+
+				// Event triggered when speech is recognized
+				speechRecognition.onresult = async (event) => {
+					// Clear the inactivity timeout
+					clearTimeout(timeoutId);
+
+					// Handle recognized speech
+					console.log(event);
+					const transcript = event.results[Object.keys(event.results).length - 1][0].transcript;
+
+					transcription = `${transcription}${transcript}`;
+
+					await tick();
+					document.getElementById('chat-textarea')?.focus();
+
+					// Restart the inactivity timeout
+					timeoutId = setTimeout(() => {
+						console.log('Speech recognition turned off due to inactivity.');
+						speechRecognition.stop();
+					}, inactivityTimeout);
+				};
+
+				// Event triggered when recognition is ended
+				speechRecognition.onend = function () {
+					// Restart recognition after it ends
+					console.log('recognition ended');
+
+					confirmRecording();
+					dispatch('confirm', transcription);
+
+					confirmed = false;
+					loading = false;
+				};
+
+				// Event triggered when an error occurs
+				speechRecognition.onerror = function (event) {
+					console.log(event);
+					toast.error($i18n.t(`Speech recognition error: {{error}}`, { error: event.error }));
+					dispatch('cancel');
+
+					stopRecording();
+				};
+			}
+		}
 	};
 
 	const stopRecording = async () => {

src/lib/components/chat/Settings/Audio.svelte

@@ -190,7 +190,7 @@
 
 			<div class=" py-0.5 flex w-full justify-between">
 				<div class=" self-center text-xs font-medium">
-					{$i18n.t('Auto-send input after 3 sec.')}
+					{$i18n.t('Instant Auto-Send After Voice Transcription')}
 				</div>
 
 				<button