/**
 * STT (Speech-to-Text) module — Whisper-compatible file-upload providers.
 *
 * Ported from src-tauri/src/stt/whisper_compat.rs
 *
 * Accepts any Whisper-compatible transcription API. All provider details
 * (endpoint, model, extra fields) are specified by the caller — nothing is
 * hard-coded here.
 *
 * Common endpoints:
 *   OpenAI Whisper  → https://api.openai.com/v1/audio/transcriptions         (model: whisper-1)
 *   Groq Whisper    → https://api.groq.com/openai/v1/audio/transcriptions     (model: whisper-large-v3-turbo)
 *   GLM-ASR         → https://open.bigmodel.cn/api/paas/v4/audio/transcriptions (model: glm-asr-2512, extraFields: {stream:'false'})
 *   SiliconFlow     → https://api.siliconflow.cn/v1/audio/transcriptions      (model: FunAudioLLM/SenseVoiceSmall)
 *
 * Accepts any audio format the provider supports: mp3, wav, m4a, webm, flac, ogg, etc.
 * No local audio processing — raw audio bytes are uploaded directly.
 *
 * Requires Node.js >= 18 (native fetch + FormData + Blob).
 */
export interface SttConfig {
    /** Full transcription API endpoint URL */
    endpoint: string;
    /** Model name to request (e.g. 'whisper-1', 'whisper-large-v3-turbo') */
    model: string;
    /** Provider API key */
    apiKey: string;
    /**
     * BCP-47 language code hint (e.g. 'zh', 'en', 'ja').
     * Omit or set to 'multi' for automatic detection.
     */
    language?: string;
    /**
     * Extra multipart form fields required by some providers.
     * e.g. GLM-ASR needs `{ stream: 'false' }`
     */
    extraFields?: Record<string, string>;
    /** Request timeout in milliseconds. Default: 60_000 */
    timeoutMs?: number;
}
/**
 * Custom STT adapter function.
 *
 * Use this when the built-in Whisper-compatible file-upload protocol doesn't
 * fit your provider (e.g. chat/completions-based audio APIs, WebSocket, gRPC…).
 *
 * @param audio    - File path or raw audio bytes (Buffer / Uint8Array)
 * @param filename - Filename hint for MIME-type inference (e.g. 'recording.m4a')
 * @returns        - Raw transcript text
 *
 * @example
 * const sdk = new VoiceTextSDK({
 *   stt: async (audio, filename) => {
 *     // your own HTTP call here
 *     return myCustomTranscribe(audio, filename);
 *   },
 *   llm: { ... },
 * });
 */
export type SttAdapter = (audio: string | Buffer | Uint8Array, filename?: string) => Promise<string>;
/**
 * Transcribe audio using a Whisper-compatible STT API.
 *
 * @param audio - Path to audio file (string) or raw audio data (Buffer/Uint8Array)
 * @param config - STT provider configuration
 * @param filename - Filename hint when `audio` is a Buffer. Used to infer MIME type.
 *                   Defaults to `'audio.wav'`.
 * @returns Raw transcript text from the provider
 *
 * @throws If the API returns an error or an empty transcript
 *
 * @example
 * // From file path
 * const text = await transcribeAudio('/path/to/recording.m4a', {
 *   provider: 'openai-whisper',
 *   apiKey: process.env.OPENAI_API_KEY!,
 *   language: 'zh',
 * });
 *
 * @example
 * // From Buffer
 * const audioBuffer = await readFile('recording.mp3');
 * const text = await transcribeAudio(audioBuffer, {
 *   provider: 'groq-whisper',
 *   apiKey: process.env.GROQ_API_KEY!,
 * }, 'recording.mp3');
 */
export declare function transcribeAudio(audio: string | Buffer | Uint8Array, config: SttConfig, filename?: string): Promise<string>;
//# sourceMappingURL=stt.d.ts.map