Asset Studio keeps reusable audio and files in your business workspace. Use it to prepare a reception greeting, record a short announcement, or generate speech from a script before adding the finished audio to a call flow.
For example, a shop can keep separate greetings for its normal opening hours and holiday closure, with names that make the correct recording easy to find.
Before you begin
Open Agents & AI → Asset Studio. Your role needs access to view the library and permission to manage assets to upload, record, generate, convert or delete them. Ask a workspace administrator if the relevant controls are unavailable.
For a microphone recording, allow microphone access in your browser. For generated speech, an enabled model and compatible voice must be available. Prepare the words first, especially business names, numbers and instructions callers need to understand.
Upload an existing file
- Select Upload files and choose the file. The upload limit is 50 MB per file.
- Wait for processing to finish. An accepted upload may still be processing; audio playback and conversion become available when the asset is ready.
- Use Edit on the uploaded item to give it a recognizable name and category. A name such as
Reception greeting — English — Septemberis easier to reuse than a recording filename. - Play the audio and check that it starts promptly, sounds clear and ends cleanly.
Categories help separate greetings, announcements and other business files. Search and the list or grid view help locate an item later. Deleting a category moves its items to Uncategorized; it does not delete those assets.
Record a greeting in the browser
Select Record, enter an asset name, and use Start recording. Speak at a steady pace, then select Stop recording and Save recording. Once the asset is ready, play it from the library to check the result.
Record in a quiet place and leave a brief pause at the beginning and end. If the browser cannot capture audio, check its microphone permission and selected input device, then try again. Saving starts processing, so confirm the resulting asset is ready before using it.
Generate speech from a script
Select Generate audio to open Generate audio with AI. Enter an asset name and Text to speak, then choose an available provider, model and voice. Scripts can contain up to 5,000 characters.
A short reception script might be:
Welcome to our shop. Press 1 for sales or 2 for support.
Select Generate audio and wait for the asset to finish processing. Listen for pronunciation, pacing and clarity. If the voice misreads a name or abbreviation, adjust the wording and generate a new version.
The spoken menu and the actual call-flow choices must agree. Mentioning “press 1” in a recording does not create that routing rule; configure and test the corresponding menu separately.
Convert and use the finished audio
For a ready audio item, Convert creates another asset in WAV or MP3 format. Choose the sample rate, channels and, for MP3, bitrate required by its destination. The original remains available while the converted copy processes.
Use Download to obtain a file, or select the prepared audio from Asset Studio where a call-flow editor offers that option. Follow the IVR and ring group guide to place it in a tested call journey.
Keep the library dependable
If an asset remains processing or fails, inspect its displayed status and try a supported source file or an available model and voice. Check existing entries before submitting the same generation repeatedly.
Before deleting an asset, check whether a call flow uses it: deleting an item also removes its linked IVR asset. Keep a known working greeting until its replacement has been tested. Open Asset Studio to prepare your next recording.
Check the greeting where customers will hear it
Listen on a phone or headset similar to the one customers will use. A recording that sounds clear through laptop speakers may have a quiet business name, a clipped first word or menu choices spoken too quickly. Ask someone who has not read the script to repeat the choices they heard.
For a bilingual menu, record each language deliberately and check pronunciation with a fluent speaker. Keep the spoken key choices consistent across the recordings and the IVR configuration. Avoid making callers wait through a long introduction before learning what to press.
Name a replacement so the team can distinguish it from the current greeting. Test it in the intended flow before removing the older asset. If you change the wording from “Press 2 for support” to another choice, update and test the routing as part of the same change.
Keep the original script with your operational notes. It is easier to correct one sentence or produce another language when the approved wording is available.