TheraVoice | Devpost

Inspiration

Mental health tools should be accessible to everyone, but many face barriers to traditional therapy. We wanted to address this gap by creating a real-time, AI-powered solution that could offer personalized, empathetic support in moments of need. The idea was to simulate therapeutic conversations, giving users thoughtful, meaningful responses whenever they sought help.

What it does

TheraVoice, our AI therapist agent built with aiXplain, takes in user input, processes it using advanced AI and NLP, and delivers a vocalized response through text-to-speech (TTS) technology. The goal was to craft a conversation that feels genuine and emotionally supportive, helping users navigate their thoughts and emotions in a way that feels personalized and conversational.

How we built it

We developed the project as a Flask app, using aiXplain’s AI agents to analyze user input and generate contextually appropriate replies. These text-based responses were then converted into a natural-sounding voice using AI/ML-powered TTS APIs. The result is an AI-driven pipeline that mimics the back-and-forth of a human conversation, where the AI listens, processes, and speaks back to the user.

Challenges we ran into

One of our biggest challenges was connecting the various components into a fully cohesive system. Integrating AI analysis with an interactive front-end, and ensuring the responses were both natural and vocally clear, proved to be more difficult than anticipated. Network issues also disrupted the TTS integration, preventing us from delivering a fully functional demo. Despite these setbacks, we’re proud of the overall progress and are excited by the possibilities this project unlocked.

Accomplishments that we're proud of

While the project didn’t reach full completion, we’re proud of the end-to-end AI pipeline we built. From understanding user input to generating empathetic responses and converting them to speech, we succeeded in creating a system that demonstrates the potential of AI-driven therapy. We also designed a clean and intuitive front-end interface for engaging with the AI agent.

What we learned

This project gave us hands-on experience with AI agents, natural language processing, and the technical challenges of building a seamless pipeline. We learned a lot about balancing AI’s technical capabilities with the need for emotional depth and support. Integrating various AI technologies into a unified system taught us valuable lessons in troubleshooting, adaptability, and design.

What's next for TheraVoice

Our immediate goal is to iron out the technical issues and ensure smooth functionality. Once stabilized, we plan to train the AI on specific therapeutic techniques to improve the depth of its responses. Enhancing the quality of the TTS system for more natural interaction and expanding the platform to support multiple languages are key milestones on our roadmap. Ultimately, we envision a versatile AI therapist that’s accessible to even more people, no matter where they are or what language they speak.