Join us in San Francisco on the 29/30th of October for two days of developer workshops and technical talks
← Back to Extend Catalog
Answering Machine detection with machine learning.

Answering Machine detection with machine learning.

Building a VAPI application that sends Text-To-Speech when an answering machine is detected.
  • Machine Learning
  • Python


For this solution, we built a machine learning algorithm that is able to detect when a call goes to voicemail by listening to the beep sound with 96% accuracy. When the call is picked up by the answering machine, we perform a speech-to-text action(TTS) which is recorded by the answering machine.

Try it out on Heroku


After you deploy the application to Heroku, make a call to the purchased number for the application.

The application will ask to enter a phone number. Enter any phone number you like, as long as it is picked up by voicemail. The call will go to voicemail and the answering machine model will start listenting on the call.

When a beep is detected, the application performs a Text-To-Speech, with the phrase, Answering Machine Detected, and the call will hangup.

To install

Clone the github repo and run:

pip install -r requirements.txt


You will need to create a Nexmo Application and Purchase a phone number



This open source project is supported by the Nexmo DevRel team on a best effort basis, issues should be raised in the GitHub repository.