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

# Use prepared audio

> Add recordings for parts of a workflow that should use a prepared message.

Use prepared audio for a greeting, a repeated message, or another part of the conversation where you want a specific delivery. Keep a voice connection configured for responses that need generated speech.

## Add audio to your workspace

Open **Audio Library** and upload a recording or use the browser recording option. Check the audio and its transcription, then give it a name you can recognize while editing an agent.

Use recordings you are authorized to use. Keep clips short and record in a quiet environment so changes between prepared and generated speech are less distracting.

## Use the recording in an agent

Open the workflow and use the prompt editor's `@` picker to select a recording where supported. Review the available speech options on greetings and transitions when you need a fixed clip.

If you want the generated speech to match a recorded speaker, configure an appropriate voice ID from your voice provider. Voice creation and any associated permissions or charges are managed by that provider.

## Test the result

Test the prepared message, a response that requires generated speech, and an interruption. Review the conversation recording to check volume, pronunciation, and transitions before publishing.