Global TTS interrupt suppression (its being a bitch)

package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@ztimson/utils",
-	"version": "0.28.9",
+	"version": "0.28.12",
 	"description": "Utility library",
 	"author": "Zak Timson",
 	"license": "MIT",

src/tts.ts

@@ -2,8 +2,11 @@ import {removeEmojis} from './string.ts';
 
 export class TTS {
 	private static readonly QUALITY_PATTERNS = ['Google', 'Microsoft', 'Samantha', 'Premium', 'Natural', 'Neural'];
+	private static _errorHandlerInstalled = false;
 
 	private _currentUtterance: SpeechSynthesisUtterance | null = null;
+	private _voicesLoaded: Promise<void>;
+	private _stoppedUtterances = new WeakSet<SpeechSynthesisUtterance>();
 
 	private _rate: number = 1.0;
 	get rate(): number { return this._rate; }
@@ -33,8 +36,9 @@ export class TTS {
 		if(this._currentUtterance && value) this._currentUtterance.voice = value;
 	}
 
-	/** Create a TTS instance with optional configuration */
 	constructor(config?: {rate?: number; pitch?: number; volume?: number; voice?: SpeechSynthesisVoice | null}) {
+		TTS.installErrorHandler();
+		this._voicesLoaded = this.initializeVoices();
 		if(config) {
 			if(config.rate !== undefined) this._rate = config.rate;
 			if(config.pitch !== undefined) this._pitch = config.pitch;
@@ -43,11 +47,31 @@ export class TTS {
 		}
 	}
 
-	/**
+	private static installErrorHandler(): void {
-	 * Selects the best available TTS voice, prioritizing high-quality options
+		if(this._errorHandlerInstalled) return;
-	 * @param lang Speaking language
+		window.addEventListener('unhandledrejection', (event) => {
-	 * @returns Highest quality voice
+			if(event.reason?.error === 'interrupted' && event.reason instanceof SpeechSynthesisErrorEvent) event.preventDefault();
-	 */
+		});
+		this._errorHandlerInstalled = true;
+	}
+
+	private initializeVoices(): Promise<void> {
+		return new Promise((resolve) => {
+			const voices = window.speechSynthesis.getVoices();
+			if(voices.length > 0) {
+				if(!this._voice) this._voice = TTS.bestVoice();
+				resolve();
+			} else {
+				const handler = () => {
+					window.speechSynthesis.removeEventListener('voiceschanged', handler);
+					if(!this._voice) this._voice = TTS.bestVoice();
+					resolve();
+				};
+				window.speechSynthesis.addEventListener('voiceschanged', handler);
+			}
+		});
+	}
+
 	private static bestVoice(lang = 'en'): SpeechSynthesisVoice | undefined {
 		const voices = window.speechSynthesis.getVoices();
 		for (const pattern of this.QUALITY_PATTERNS) {
@@ -57,14 +81,12 @@ export class TTS {
 		return voices.find(v => v.lang.startsWith(lang));
 	}
 
-	/** Cleans text for TTS by removing emojis, markdown and code block */
 	private static cleanText(text: string): string {
 		return removeEmojis(text)
 			.replace(/```[\s\S]*?```/g, ' code block ')
 			.replace(/[#*_~`]/g, '');
 	}
 
-	/** Creates a speech utterance with current options */
 	private createUtterance(text: string): SpeechSynthesisUtterance {
 		const cleanedText = TTS.cleanText(text);
 		const utterance = new SpeechSynthesisUtterance(cleanedText);
@@ -76,63 +98,50 @@ export class TTS {
 		return utterance;
 	}
 
-	/** Speaks text and returns a Promise which resolves once complete */
+	async speak(text: string): Promise<void> {
-	speak(text: string): Promise<void> {
 		if(!text.trim()) return Promise.resolve();
-
+		await this._voicesLoaded;
 		return new Promise((resolve, reject) => {
 			this._currentUtterance = this.createUtterance(text);
-
+			const utterance = this._currentUtterance;
-			this._currentUtterance.onend = () => {
+			utterance.onend = () => {
 				this._currentUtterance = null;
 				resolve();
 			};
-
+			utterance.onerror = (error) => {
-			this._currentUtterance.onerror = (error) => {
 				this._currentUtterance = null;
-				reject(error);
+				if(this._stoppedUtterances.has(utterance) && error.error === 'interrupted') resolve();
+				else reject(error);
 			};
-
+			window.speechSynthesis.speak(utterance);
-			window.speechSynthesis.speak(this._currentUtterance);
 		});
 	}
 
-	/** Stops all TTS */
 	stop(): void {
+		if(this._currentUtterance) this._stoppedUtterances.add(this._currentUtterance);
 		window.speechSynthesis.cancel();
 		this._currentUtterance = null;
 	}
 
-	/**
+	speakStream(): {next: (text: string) => void, done: () => Promise<void>} {
-	 * Initialize a stream that chunks text into sentences and speak them.
-	 *
-	 * @example
-	 * const stream = tts.speakStream();
-	 * stream.next("Hello ");
-	 * stream.next("World. How");
-	 * stream.next(" are you?");
-	 * await stream.done();
-	 *
-	 * @returns Object with next function for passing chunk of streamed text and done for completing the stream
-	 */
-	speakStream(): {next: (text: string) => void, done: () => void} {
 		let buffer = '';
+		let streamPromise: Promise<void> = Promise.resolve();
 		const sentenceRegex = /[^.!?\n]+[.!?\n]+/g;
-
 		return {
 			next: (text: string): void => {
 				buffer += text;
 				const sentences = buffer.match(sentenceRegex);
 				if(sentences) {
-					sentences.forEach(sentence => this.speak(sentence.trim()));
+					sentences.forEach(sentence => streamPromise = this.speak(sentence.trim()));
 					buffer = buffer.replace(sentenceRegex, '');
 				}
 			},
 			done: async (): Promise<void> => {
 				if(buffer.trim()) {
-					await this.speak(buffer.trim());
+					streamPromise = this.speak(buffer.trim());
 					buffer = '';
 				}
+				await streamPromise;
 			}
 		};
 	}