export declare const ASR_OUTPUT_FRAME_SAMPLES = 320;
export interface PhonemeSpan {
    phoneme: string;
    startMs: number;
    endMs: number;
    startFrame: number;
    endFrame: number;
}
export interface CTCSpanOptions {
    blankId: number;
    wordBoundaryToken: string;
    frameMs: number;
    collapseRepeats?: boolean;
}
/** One PhonemeSpan per emitted phoneme (same emissions as decodeCtcTokens, `<wb>`
 *  excluded), tiling the output-frame timeline: a phoneme runs from the frame it is
 *  emitted until the next emission (or the end). */
export declare const ctcPhonemeSpans: (frameTokenIds: readonly number[], decoderTokens: readonly string[], options: CTCSpanOptions) => PhonemeSpan[];