videosphere
    Preparing search index...
    • Takes a caption-sized chunk off oversized transcript when possible. Once past softMax, prefers the last complete sentence that still fits in that window so continuous speech becomes short caption/TTS units.

      Parameters

      • committed: string

        Transcript awaiting a pause / provider final.

      • softMax: number = STREAMING_UTTERANCE_SOFT_MAX_CHARS

        Prefer a break once length reaches this.

      • hardMax: number = STREAMING_UTTERANCE_HARD_MAX_CHARS

        Force a clause/word break at/before this length when needed.

      • options: TakeUtteranceChunkOptions = {}

        Partial vs final splitting behavior.

      Returns { chunk: string; rest: string }

      Chunk to emit as final plus remaining text, or null to wait.