VidStudio SDK reference
    Preparing search index...

    Interface TranscriptSegment

    One spoken paragraph (or word, post task 11) on the transcript, as persisted.

    This is the public TranscriptSegment (what getTranscript() returns) plus two storage-only fields. Extending the public type keeps the two shapes from drifting: anything the frontend sees round-trips through IDB unchanged, and the conversion back to the public shape (task 09) is just dropping the FK and order.

    interface TranscriptSegment {
        endMs: number;
        id: string;
        kind: "speech" | "silence" | "word";
        order: number;
        parent_id?: string;
        speakerId?: string;
        startMs: number;
        text: string;
        transcript_record_id: string;
        words?: WordToken[];
    }

    Hierarchy

    • TranscriptSegment
      • TranscriptSegment
    Index
    endMs id kind order parent_id? speakerId? startMs text transcript_record_id words?
    endMs: number
    id: string
    kind: "speech" | "silence" | "word"
    order: number
    parent_id?: string

    For kind: 'word' rows only: the id of the parent kind: 'speech' paragraph this word belongs to. Undefined on speech/silence rows.

    speakerId?: string
    startMs: number

    text

    text: string
    transcript_record_id: string
    words?: WordToken[]

    Word-level timing for a speech paragraph, populated when a word-timestamped model produced the transcript. Present on kind: 'speech' segments returned by getTranscript(); absent on silences and on transcripts made by segment-only models. Used for word-accurate cuts (snap a selection to word boundaries). Word rows are never returned as top-level segments — they are nested here under their parent paragraph.

    Settings
    Member Visibility

    On This Page
    Properties