ReqLLM.Transcription.Result (ReqLLM v1.29.0)

Copy Markdown View Source

Result of an audio transcription operation.

Contains the transcribed text, optional timing segments, detected language, and audio duration. Inspired by the Vercel AI SDK's TranscriptionResult.

Fields

  • :text - The complete transcribed text from the audio input
  • :segments - List of transcript segments with timing information. Each segment is a map with:
    • :text - The text content of this segment
    • :start_second - Start time in seconds
    • :end_second - End time in seconds
  • :language - The detected language in ISO-639-1 format (e.g., "en"), or nil
  • :duration_in_seconds - The total duration of the audio in seconds, or nil

Examples

%ReqLLM.Transcription.Result{
  text: "Hello, how are you?",
  segments: [
    %{text: "Hello,", start_second: 0.0, end_second: 0.5},
    %{text: " how are you?", start_second: 0.5, end_second: 1.2}
  ],
  language: "en",
  duration_in_seconds: 1.2
}

Summary

Types

segment()
t()

Types

segment()

@type segment() :: %{text: String.t(), start_second: float(), end_second: float()}

t()

@type t() :: %ReqLLM.Transcription.Result{
  duration_in_seconds: float() | nil,
  language: String.t() | nil,
  segments: [segment()],
  text: String.t()
}