Adaptive AI VoiceTrainer:
Enhancing Sales Training Efficiency with Adaptive AI Interactive Voice Response System
Project summary
The adaptive AI phone application, tailored for KLM, leverages state-of-the-art artificial intelligence to elevate the training and performance of sales representatives. Adopting a systems engineering approach, I meticulously crafted the software's functional architecture. To set the stage, I developed an abstract usage scenario, wherein I took on the role of the principal actor to vividly depict this scenario in a short film:
Drawing inspiration from the abstract usage scenario portrayed in the video, specific usage scenarios were elaborated upon, categorized into deployment and operational phases.
The deployment phase encompasses the initial setup of the application, incorporating customer data and server initialization. The operational phase, on the other hand, is dedicated to managing actual phone calls. Critical functionalities during this phase include integration with voice over the internet protocol, capturing voice inputs, converting speech to text, generating responses through AI, and transforming text back into speech. The architecture of the software is systematically illustrated via a functional hierarchy tree and a comprehensive activity diagram:
Throughout the iterative development process, the focus was on enhancing the software's capabilities in real-time. This involved keen attention to minimizing latency and improving voice quality. The final iteration of the product is a seamless amalgamation of several cutting-edge technologies. It integrates Vosk for robust speech recognition, ElevenLabs for lifelike text-to-speech conversion, ChatGPT 4 for nuanced language processing, and Twilio’s Voice API to transform the application into a fully-functional VOIP phone system.
This all-encompassing system is specifically designed to revolutionize the training experience for sales representatives, enabling them to adeptly manage the unpredictability of real-life customer interactions, thereby bolstering their confidence and effectiveness.
Project documents
The report:
The presentation:
The poster:
Product
The product is a mobile application that operates with a live server. Anyone who has the application's phone number can test it out. If you would like to give it a try, please send me an email, and we can configure the server for you to experiment with.
Code implementation
Hidden for now