Step 1: Create a Voice Clone
You need a voice clone before you can use audio variables.
Go to Voice Models in the side navigation and click Create new voice. Name your voice and choose its language. Then choose how to provide audio:
Record – Record 4–10 minutes in a quiet room for best results.
Upload – Upload .mp3 or .mp4 files (under 11MB total).
Clone from library – Use an existing webcam video from your library.
Click Create voice to process your clone.
Step 2: Record Your Video
Record your webcam video as usual. Add a short pause before and after any words you plan to turn into audio variables. This makes the AI-generated replacements sound more natural.
Step 3: Add Audio Variables in the Video Creator
Open your campaign and go to the Video Creator. In the right sidebar, open the Audio Variables tab. You’ll see the transcript of your video.
Create a variable: Click a word in the transcript. In the popup, choose Make audio variable.
Configure it: In the modal you can:
Edit the content – Replace the text with variables like {{First Name}} or {{Company}}.
Adjust the selection – Drag the handles on the waveform to choose exactly which part of the audio is replaced.
Greeting rule: For variables in the first 4 seconds, MailMoo includes all words from the start. This improves quality for greetings like “Hi {{First Name}}.”
Ensure that you've set the audio variable's placement correctly. The audio variable segment should start when your speech starts and last slightly before the next speech starts.
For the best results, ensure that you've left a brief pause before and after the word you intend on making the audio variable.
Step 4: Select Your Voice Model
In the campaign or video creator, use the Voice Model selector and choose the voice clone you created. That voice is used for all audio variables in the campaign.
Tips
Record 4-10 minutes of audio for your voice clone.
Add pauses around words you’ll turn into variables. (very important)
Keep variables short (names, company names, etc.).
Use the same language as your voice clone for best results.
Troubleshooting
Sounds unnatural: Adjust the start/end handles on the waveform and keep pauses around the variable.
Wrong transcript: Click a word and choose Correct, or use Retranscribe video to regenerate the transcript.
No personalization: Ensure you use {{Variable Name}} placeholders and that the names match your lead data columns.