Documentation / Audio library
Audio library
Upload or record audio clips for use in agent responses.
The library contains prepared audio for the agent to use. It differs from conversation recordings available in runs.
Add audio
You need workspace access and permission to create agent resources.
- Open the recordings/audio library area.
- Use the add-audio action.
- Select files or use the recording controls offered by the form.
- Check the audio's language and details before uploading.
- Wait for completion and inspect the item in the list.
Use names that identify purpose, such as a greeting or opening-hours notice. Listen before attaching audio to avoid empty clips or instructions for another scenario.
Make it available to the agent
Open the agent's recordings configuration and select relevant clips. Review instructions explaining when to use prepared responses. Save and test a question that matches the clip.
A library file does not mean every agent step should use it. Check its selection and behavior in a run.
Verify the experience
Check volume, leading silence, language and conversation continuity. Compare with the agent's other speech and test interruptions when enabled. If a prepared response does not fit the situation, review the selection and instructions.
When replacing or removing audio, review dependent agents and test again before publishing.
For audio from a completed call, follow Runs and usage, not library upload.