> ## Documentation Index
> Fetch the complete documentation index at: https://docs.youniverse.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Transcribe audio and video

> Turn recorded speech into text with language, speaker and audio-event options.

<Frame>
  <img src="https://chat.youniverse.ai/__l5e/assets-v1/982b9812-45cc-4e3a-be1d-6432d2bfceea/docs-studio-speech.jpg" alt="Glass microphone beside a transcription script" />
</Frame>

Open **AI Studio → Transcribe** to convert an audio or video recording into text. This is a file-based transcription workspace, not the live voice-chat microphone.

## Upload and configure

<Steps>
  <Step title="Choose the recording">Drop an audio/video file into the upload area or select it from your device. The upload ceiling is 1 GB.</Step>
  <Step title="Choose language">Use Auto-detect or select the recording's language when you know it.</Step>
  <Step title="Choose speaker and event options">Speaker diarization separates different speakers. Audio event tagging includes supported non-speech events such as laughter or music.</Step>
  <Step title="Transcribe and inspect">Run transcription, then review the text against the recording before using it as a final record.</Step>
</Steps>

## What the options do

| Option | Use it when |
| - | - |
| Auto-detect | You are unsure of the spoken language. |
| Manual language | You know the language and want to specify it. |
| Speaker diarization | A meeting, interview or recording contains multiple speakers. |
| Audio event tagging | Non-speech sounds matter to the transcript. |

Speaker labels distinguish voices; they do not reliably identify a person's real name. Correct names, technical terms and numbers during review.

<Frame>
  <img src="https://chat.youniverse.ai/__l5e/assets-v1/074267d1-c0dd-4ba2-8221-041306318960/docs-studio-music-editing.jpg" alt="Separate glass audio segments representing speaker turns" />
</Frame>

## Copy or download the transcript

Use **Copy transcript** to copy the result, or **Download transcript** to save a `.txt` file with timestamps and speaker labels. Review the transcription before treating those labels and words as final.

## Better source material

Use a clear recording with minimal background noise. Speaker overlap, distant microphones and music can make transcription harder. If the result is important, listen and check it instead of relying on an unreviewed transcript.

<AccordionGroup>
  <Accordion title="Can I upload a video?">Yes. Transcribe accepts video as well as audio and works on its spoken content.</Accordion>
  <Accordion title="Can I transcribe a live conversation here?">This tab works on an uploaded recording. Use [voice chat](/chat/voice) for a live conversation with Iris.</Accordion>
  <Accordion title="Is my recording covered by ZDR?">The Transcribe workspace displays a Zero Data Retention badge. This does not extend to unrelated music or video generation.</Accordion>
</AccordionGroup>

<Info>Transcription controls require Pro or Elite. Use recordings you have permission to process.</Info>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.