> ## Documentation Index
> Fetch the complete documentation index at: https://hyperframes-codex-docs-human-first-refactor.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Transcribe media and create captions

> Turn speech into editable text, synchronize it, and finish the captions in Studio.

Transcription creates timed words from audio or video. Captions turn those words into a readable visual layer.

## A practical workflow

<Steps>
  <Step title="Transcribe the media">
    Ask the agent to transcribe the source, or use the CLI:

    ```bash theme={null}
    npx hyperframes transcribe interview.mp4
    ```
  </Step>

  <Step title="Correct the text">
    Fix names, product terms, numbers, punctuation, and obvious recognition errors before styling.
  </Step>

  <Step title="Create caption segments">
    Group words into short readable phrases while keeping their timing connected to the speech.
  </Step>

  <Step title="Edit in Studio">
    Open [Captions](/studio/captions) to adjust text, timing, placement, style, and animation.
  </Step>

  <Step title="Review twice">
    Watch once with sound for synchronization and once muted for readability.
  </Step>
</Steps>

## Good caption decisions

* Keep lines short enough to scan.
* Break on meaning rather than an arbitrary word count.
* Do not cover faces or product controls.
* Use emphasis sparingly.
* Keep contrast reliable as the background changes.
* Preserve the exact spoken meaning.

Use [Captions or recuts](/guides/captions-and-recuts) when starting from existing talking-head footage.