js-tts-wrapper/engines/sherpaonnx.d.ts

A JavaScript/TypeScript library that provides a unified API for working with multiple cloud-based Text-to-Speech (TTS) services

import { AbstractTTSClient } from "../core/abstract-tts";
import type { SpeakOptions, TTSCredentials, UnifiedVoice } from "../types";
/**
 * SherpaOnnx TTS credentials
 */
export interface SherpaOnnxTTSCredentials extends TTSCredentials {
    /**
     * Path to model file or directory
     */
    modelPath?: string;
    /**
     * Path to tokens file
     */
    tokensPath?: string;
    /**
     * Voice model ID
     */
    modelId?: string;
    /**
     * If true, skip automatic download of default model
     */
    noDefaultDownload?: boolean;
}
/**
 * SherpaOnnx TTS client
 */
export declare class SherpaOnnxTTSClient extends AbstractTTSClient {
    /**
     * Path to the model file
     */
    private modelPath;
    /**
     * Voice model ID
     */
    private modelId;
    /**
     * Base directory for models
     */
    private baseDir;
    /**
     * SherpaOnnx TTS instance
     */
    private tts;
    /**
     * Sample rate
     */
    private sampleRate;
    /**
     * Model configuration
     */
    private jsonModels;
    /**
     * Path to the models file
     */
    private static readonly MODELS_FILE;
    /**
     * Create a new SherpaOnnx TTS client
     * @param credentials SherpaOnnx credentials
     */
    constructor(credentials: SherpaOnnxTTSCredentials);
    /**
     * Load models and voices from the JSON configuration file
     * @returns Record of model configurations
     */
    private loadModelsAndVoices;
    /**
     * Download the models file from the repository
     */
    private downloadModelsFile;
    /**
     * Download a file from a URL to a destination path
     * @param url URL to download from
     * @param destination Destination path
     * @returns Promise resolving when the download is complete
     */
    private downloadFile;
    /**
     * Extract a tar.bz2 archive to a destination directory
     * @param archivePath Path to the archive file
     * @param destinationDir Destination directory
     * @returns Promise resolving to a map of extracted file paths
     */
    private extractTarBz2;
    /**
     * Check if model and token files exist
     * @param modelPath Path to model file
     * @param tokensPath Path to tokens file
     * @returns True if both files exist and are not empty
     */
    private checkFilesExist;
    /**
     * Get dict_dir from extracted model
     * @param destinationDir Destination directory
     * @returns Path to dict_dir
     */
    private getDictDir;
    /**
     * Download model and token files to voice-specific directory
     * @param destinationDir Base directory for model files
     * @param modelId Voice model ID
     * @returns Tuple of (model_path, tokens_path, lexicon_path, dict_dir)
     */
    private downloadModelAndTokens;
    /**
     * Check if model exists and download if not
     * @param modelId Voice model ID
     * @returns Tuple of (model_path, tokens_path, lexicon_path, dict_dir)
     */
    private checkAndDownloadModel;
    /**
     * Initialize the SherpaOnnx TTS engine
     * @param modelPath Path to model file
     * @param tokensPath Path to tokens file
     */
    private initializeTTS;
    /**
     * Estimate word boundaries from text
     * @param text Text to estimate word boundaries for
     * @param audioDuration Duration of the audio in seconds
     * @returns Array of word boundaries
     */
    private estimateWordBoundaries;
    /**
     * Get available voices from the provider
     * @returns Promise resolving to an array of voice objects
     */
    protected _getVoices(): Promise<any[]>;
    /**
     * Map SherpaOnnx voice objects to unified format
     * @param rawVoices Array of SherpaOnnx voice objects
     * @returns Promise resolving to an array of unified voice objects
     */
    protected _mapVoicesToUnified(rawVoices: any[]): Promise<UnifiedVoice[]>;
    /**
     * Set the voice to use for synthesis
     * @param voiceId Voice ID to use
     */
    setVoice(voiceId: string): Promise<void>;
    /**
     * Convert text to audio bytes
     * @param text Text to synthesize
     * @returns Promise resolving to audio bytes
     */
    synthToBytes(text: string): Promise<Uint8Array>;
    /**
     * Synthesize text to a byte stream with word boundaries
     * @param text Text to synthesize
     * @param options Synthesis options
     * @returns Promise resolving to a readable stream of audio bytes with word boundaries
     */
    synthToBytestream(text: string, options?: SpeakOptions): Promise<ReadableStream<Uint8Array> | {
        audioStream: ReadableStream<Uint8Array>;
        wordBoundaries: Array<{
            text: string;
            offset: number;
            duration: number;
        }>;
    }>;
    /**
     * Strip SSML tags from text
     * @param text Text with SSML tags
     * @returns Plain text without SSML tags
     */
    private stripSSML;
    /**
     * Check if credentials are valid
     * @returns Promise resolving to true if credentials are valid
     */
    checkCredentials(): Promise<boolean>;
}