How we created human-like interactions through fast AI prototyping
Imagine800 is a Spain-based technology company specializing in innovative telephone communication solutions. Their clients handle over 50,000 calls daily, but operators often spend valuable time on unanswered or prematurely ended calls.
Our goal was to maximize the efficiency of the client’s operators by developing an AI-driven calling system with natural voice synthesis.
The system needed to manage the initial stages of each call, delivering a pre-recorded message upon a successful connection. Additionally, the generated voice had to sound natural and human-like to ensure a seamless interaction.
We developed an AI-driven system capable of initiating multiple calls simultaneously. Each call begins with the AI handling the first interaction, greeting the receiver with a natural, human-like voice. If the receiver responds, the system seamlessly transitions to delivering a pre-recorded message.
Our focus was on developing a rapid proof of concept within a one-month deadline. Following research and initial testing, we selected XTTSv2, a multilingual text-to-speech tool offering higher-quality output in Spanish, making it a natural fit for the project. To enhance the synthesized voice further, we fine-tuned it for clarity, tone, and overall accuracy, ensuring an authentic and engaging experience for listeners.
We focused on refining the synthesized voice through manual adjustments, specifically fine-tuning pitch and cadence to closely match the intended tone and ensure natural sound quality. The fast prototyping process allowed us to complete the proof of concept within two development sprints, totaling just four weeks.
Once deployed, the AI-driven calling tool is designed to automate a large volume of daily calls, significantly enhancing operational efficiency.
This allows the client to focus on meaningful interactions while preserving a human touch in customer engagement.
"The relationship was very good; from the first minute, the team was open to adapting to any scenario we proposed, and they still show that positive attitude."