DictationAudioInput:
    | string
    | Uint8Array
    | ArrayBuffer
    | Blob
    | ReadableStream<Uint8Array>
    | NodeJS.ReadableStream
    | AsyncIterable<Uint8Array>
    | Iterable<Uint8Array>

Audio for a dictation request.

The dictation API opens one connection for every shape audio comes in, so this takes both: audio still being produced — an async iterable (Node streams included), a sync iterable, or a web ReadableStream of Uint8Array chunks — and audio already complete: a local file path or a data URL (file system access requires Node.js, Bun, or Deno), raw bytes, or a Blob/File, which is sent as a single chunk over that same connection.

URLs are not accepted — the dictation API has no URL ingestion; use client.transcripts for URL or asynchronous transcription.