adam | Devpost

Landing page
Recording page
During recording
History page
Loading screen
AI voice call
AI chatbot

Inspiration

We were inspired by the growing conversation around mental health and the fact that a significant portion of anxiety and stress stems from social interactions. We wanted to build a tool that acts as a supportive friend, helping people navigate these moments in real-time. Our goal was to create a technological solution that not only identifies emotional states but also empowers individuals to manage them, fostering greater self-awareness, reducing social anxiety, and promoting healthier, more authentic connections.

What it does

Adam is a speech emotion recognition (SER) tool that provides real-time conversational assistance. Using a microphone, it listens to a user's speech, analyzing the tone, cadence, and pauses. It then uses natural language processing (NLP) to understand the context of what is being said. Based on this analysis, our AI identifies the user's emotional state—such as nervousness, hesitation, or anxiety—and provides immediate, personalized solutions. These solutions are not just about finding the right words, but about providing a moment of calm and clarity, helping users to better regulate their emotions and reduce the mental burden of challenging conversations.

How we built it

One of our biggest challenges was achieving high accuracy in real-time emotion detection, especially for nuanced emotional states like subtle anxiety. We also grappled with the ethical considerations of building a tool related to mental health. It was a significant challenge to design a user experience that felt helpful and supportive, rather than intrusive or critical, and to ensure that our solutions were gentle and non-judgmental. We spent a lot of time ensuring our AI provides constructive support without overstepping into the role of a professional therapist.

Challenges we ran into

Accomplishments that we're proud of

We are most proud of building a tool that can genuinely help people in moments of social anxiety. Getting the system to not just identify an emotion but to offer genuinely useful, calming advice in the moment was a major technical and design achievement. We were also proud to create a user-friendly interface that makes this complex technology feel accessible and non-intimidating, proving that AI can be a compassionate tool for personal growth and mental wellness rather than just a cold, analytical machine.

What we learned

This project taught us a tremendous amount about the nuances of both human communication and machine learning. We learned that the context of a conversation is just as important as the vocal cues. We also gained valuable experience in fine-tuning machine learning models for low-latency, real-time applications. Most importantly, we learned the profound importance of human-centered design when building AI tools, ensuring that the technology serves the user's emotional and mental well-being first.

What's next for Adam

We have several exciting plans for the future of Adam. We want to expand our emotion recognition to include more subtle emotional states, such as boredom or engagement, which can be indicators of social distress. We also plan to develop a feature that provides post-conversation summaries, offering users a look at their emotional trends over time to help them track their progress in managing social anxiety and stress. Additionally, we're exploring the possibility of integrating Adam with video to analyze non-verbal cues, like facial expressions, to provide an even more comprehensive communication and mental wellness coach. We also plan to add a disclaimer and resources page to guide users to professional help if needed.