Added TTS helper
This commit is contained in:
@@ -21,5 +21,6 @@ export * from './search';
|
||||
export * from './string';
|
||||
export * from './template';
|
||||
export * from './time';
|
||||
export * from './tts';
|
||||
export * from './types';
|
||||
export * from 'var-persist';
|
||||
|
||||
@@ -140,7 +140,7 @@ export function pascalCase(str?: string): string {
|
||||
* @param {string} str Input string with emojis
|
||||
* @returns {string} Sanitized string without emojis
|
||||
*/
|
||||
function removeEmojis(str: string): string {
|
||||
export function removeEmojis(str: string): string {
|
||||
const emojiRegex = /(?:[\u2700-\u27bf]|(?:\ud83c[\udde6-\uddff]){2}|[\ud83c[\udde6-\uddff]|[\ud83d[\ude00-\ude4f]|[\ud83d[\ude80-\udeff]|[\ud83c[\udd00-\uddff]|[\ud83d[\ude50-\ude7f]|[\u2600-\u26ff]|[\u2700-\u27bf]|[\ud83e[\udd00-\uddff]|[\ud83c[\udf00-\uffff]|[\ud83d[\ude00-\udeff]|[\ud83c[\udde6-\uddff])/g;
|
||||
return str.replace(emojiRegex, '');
|
||||
}
|
||||
|
||||
139
src/tts.ts
Normal file
139
src/tts.ts
Normal file
@@ -0,0 +1,139 @@
|
||||
import {removeEmojis} from './string.ts';
|
||||
|
||||
export class TTS {
|
||||
private static readonly QUALITY_PATTERNS = ['Google', 'Microsoft', 'Samantha', 'Premium', 'Natural', 'Neural'];
|
||||
|
||||
private _currentUtterance: SpeechSynthesisUtterance | null = null;
|
||||
|
||||
private _rate: number = 1.0;
|
||||
get rate(): number { return this._rate; }
|
||||
set rate(value: number) {
|
||||
this._rate = value;
|
||||
if(this._currentUtterance) this._currentUtterance.rate = value;
|
||||
}
|
||||
|
||||
private _pitch: number = 1.0;
|
||||
get pitch(): number { return this._pitch; }
|
||||
set pitch(value: number) {
|
||||
this._pitch = value;
|
||||
if(this._currentUtterance) this._currentUtterance.pitch = value;
|
||||
}
|
||||
|
||||
private _volume: number = 1.0;
|
||||
get volume(): number { return this._volume; }
|
||||
set volume(value: number) {
|
||||
this._volume = value;
|
||||
if(this._currentUtterance) this._currentUtterance.volume = value;
|
||||
}
|
||||
|
||||
private _voice: SpeechSynthesisVoice | undefined;
|
||||
get voice(): SpeechSynthesisVoice | undefined { return this._voice; }
|
||||
set voice(value: SpeechSynthesisVoice | undefined) {
|
||||
this._voice = value;
|
||||
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}) {
|
||||
if(config) {
|
||||
if(config.rate !== undefined) this._rate = config.rate;
|
||||
if(config.pitch !== undefined) this._pitch = config.pitch;
|
||||
if(config.volume !== undefined) this._volume = config.volume;
|
||||
this._voice = config.voice === null ? undefined : (config.voice || undefined);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Selects the best available TTS voice, prioritizing high-quality options
|
||||
* @param lang Speaking language
|
||||
* @returns Highest quality voice
|
||||
*/
|
||||
private static bestVoice(lang = 'en'): SpeechSynthesisVoice | undefined {
|
||||
const voices = window.speechSynthesis.getVoices();
|
||||
for (const pattern of this.QUALITY_PATTERNS) {
|
||||
const voice = voices.find(v => v.name.includes(pattern) && v.lang.startsWith(lang));
|
||||
if(voice) return voice;
|
||||
}
|
||||
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);
|
||||
const voice = this._voice || TTS.bestVoice();
|
||||
if(voice) utterance.voice = voice;
|
||||
utterance.rate = this._rate;
|
||||
utterance.pitch = this._pitch;
|
||||
utterance.volume = this._volume;
|
||||
return utterance;
|
||||
}
|
||||
|
||||
/** Speaks text and returns a Promise which resolves once complete */
|
||||
speak(text: string): Promise<void> {
|
||||
if(!text.trim()) return Promise.resolve();
|
||||
|
||||
return new Promise((resolve, reject) => {
|
||||
this._currentUtterance = this.createUtterance(text);
|
||||
|
||||
this._currentUtterance.onend = () => {
|
||||
this._currentUtterance = null;
|
||||
resolve();
|
||||
};
|
||||
|
||||
this._currentUtterance.onerror = (error) => {
|
||||
this._currentUtterance = null;
|
||||
reject(error);
|
||||
};
|
||||
|
||||
window.speechSynthesis.speak(this._currentUtterance);
|
||||
});
|
||||
}
|
||||
|
||||
/** Stops all TTS */
|
||||
stop(): void {
|
||||
window.speechSynthesis.cancel();
|
||||
this._currentUtterance = null;
|
||||
}
|
||||
|
||||
/**
|
||||
* 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 = '';
|
||||
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()));
|
||||
buffer = buffer.replace(sentenceRegex, '');
|
||||
}
|
||||
},
|
||||
done: async (): Promise<void> => {
|
||||
if(buffer.trim()) {
|
||||
await this.speak(buffer.trim());
|
||||
buffer = '';
|
||||
}
|
||||
}
|
||||
};
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user