import type { TTS as TTSBase } from '../agents/providers.js';
/** Thrown when a requested provider/feature is not available in the current build. */
export declare class NotImplementedError extends Error {
    constructor(message: string);
}
/** Options for {@link TTS.cartesia} (Cartesia). */
export type TTSCartesiaOptions = {
    /** Provider identifier; normally left unset. */
    provider?: string | null;
    /** API key for the provider. */
    apiKey?: string | null;
    /** Model name. Defaults to `"sonic-2"`. */
    modelId?: string | null;
    /** Voice id or embedding; alias for {@link TTSCartesiaOptions.voice}. */
    voiceId?: string | number[] | null;
    /** Voice id or embedding. Defaults to a built-in Cartesia voice. */
    voice?: string | number[] | null;
    /** Synthesis language. Defaults to `"en"`. */
    language?: string | null;
    /** Output audio sample rate in Hz. Defaults to 24000. */
    sampleRate?: number | null;
    /** Speaking speed. */
    speed?: number | null;
    /** Output volume. */
    volume?: number | null;
    /** Emotion/style tag for synthesis. */
    emotion?: string | null;
    /** Override the inference endpoint base URL. */
    baseUrl?: string | null;
};
/** Options for {@link TTS.google} (Google). */
export type TTSGoogleOptions = {
    /** Provider identifier; normally left unset. */
    provider?: string | null;
    /** API key for the provider. */
    apiKey?: string | null;
    /** Model name. Defaults to `"Chirp3-HD"`. */
    modelId?: string | null;
    /** Voice name; alias for {@link TTSGoogleOptions.voice}. */
    voiceId?: string | null;
    /** Voice name. Defaults to `"Achernar"`. A bare name (fewer than 3 dashes) is composed into the full voice as `{language}-{model}-{voice}`. */
    voice?: string | null;
    /** Synthesis language code (BCP-47); alias for {@link TTSGoogleOptions.language}. */
    languageCode?: string | null;
    /** Synthesis language. Defaults to `"en-US"`. */
    language?: string | null;
    /** Provider region/location. */
    location?: string | null;
    /** Output audio sample rate in Hz. Defaults to 24000. */
    sampleRate?: number | null;
    /** Speaking rate multiplier. */
    speakingRate?: number | null;
    /** Pitch adjustment. */
    pitch?: number | null;
    /** Override the inference endpoint base URL. */
    baseUrl?: string | null;
};
/** Options for {@link TTS.sarvam} (Sarvam AI). */
export type TTSSarvamOptions = {
    /** Provider identifier; normally left unset. */
    provider?: string | null;
    /** API key for the provider. */
    apiKey?: string | null;
    /** Model name. Defaults to `"bulbul:v3"`. */
    modelId?: string | null;
    /** Speaker voice name. Defaults to `"shubh"`. */
    speaker?: string | null;
    /** Speaker voice name; alias for {@link TTSSarvamOptions.speaker}. */
    voiceId?: string | null;
    /** Synthesis language. Defaults to `"en-IN"`. */
    language?: string | null;
    /** Override the inference endpoint base URL. */
    baseUrl?: string | null;
};
/** Options for {@link TTS.deepgram} (Deepgram). Extra keys are passed through to the provider. */
export type TTSDeepgramOptions = {
    /** Provider identifier; normally left unset. */
    provider?: string | null;
    /** Model name; alias for {@link TTSDeepgramOptions.model}. Defaults to `"aura-2"`. */
    modelId?: string | null;
    /** Model name. Defaults to `"aura-2"`. */
    model?: string | null;
    /** Voice name. Defaults to `"asteria"`. */
    voiceId?: string | null;
    /** Synthesis language. Defaults to `"en"`. */
    language?: string | null;
    /** Output audio sample rate in Hz. Defaults to 24000. */
    sampleRate?: number | null;
    /** Audio encoding. */
    encoding?: string | null;
    /** Audio container format. */
    container?: string | null;
    /** Output bit rate. */
    bitRate?: number | null;
    /** Override the inference endpoint base URL. */
    baseUrl?: string | null;
    /** Additional provider-specific options. */
    [key: string]: any;
};
/** Factory for text-to-speech components used by the direct inference API. */
export declare class TTS {
    /** Build a Cartesia text-to-speech component. Defaults: model `"sonic-2"`, language `"en"`, sample rate 24000 Hz, plus a built-in default voice. */
    static cartesia(opts?: TTSCartesiaOptions): TTSBase;
    /** Build a Google text-to-speech component. Defaults: model `"Chirp3-HD"`, voice `"Achernar"`, language `"en-US"`, sample rate 24000 Hz. A bare voice name (fewer than 3 dashes) is composed into `{language}-{model}-{voice}`. */
    static google(opts?: TTSGoogleOptions): TTSBase;
    /** Build a Sarvam AI text-to-speech component. Defaults: model `"bulbul:v3"`, speaker `"shubh"`, language `"en-IN"`, sample rate 24000 Hz. */
    static sarvam(opts?: TTSSarvamOptions): TTSBase;
    /** Build a Deepgram text-to-speech component. Defaults: model `"aura-2"`, voice `"amalthea"`, language `"en"`, sample rate 24000 Hz. */
    static deepgram(opts?: TTSDeepgramOptions): TTSBase;
}