Mendi

Your voice. Your space. Your calm.

Comment

Inspiration

In 2021, I went through a period of major depression that nearly broke me. I felt isolated, unheard, and emotionally drained. I was unable to articulate exactly what I was going through. What helped me most wasn’t advice or fixes, but simply being heard. That experience planted a seed: what if we could give people a way to express themselves freely through their own voice and feel acknowledged, supported, and understood? This project was born from that idea: an AI-powered voice companion that listens with intention, speaks with empathy, and creates a safe space for reflection. It’s not just a tool—it’s a bridge for moments when talking to a human feels too hard, but silence is unbearable.

What it does

This application is an AI-powered voice/ text companion designed to provide emotional support. Users speak to it naturally, and it listens empathetically, responds with a realistic voice (via ElevenLabs or OpenAI), main purpose is to help users feel heard, supported, and less isolated especially during stressful or lonely times.

Challenges we ran into

Generating empathetic responses, converting them to speech to a empathatic conversational experience. Maintaining long-term conversational context and interpreting emotions into a natural speech which is currently difficult, especially without persistent memory.

Accomplishments that we're proud of

Successfully delivered a version 0.0.1 of cross-functional application that enables natural, real-time conversations with AI, combining speech recognition, emotional intelligence, and synthetic speech an ambitious technical integration. Created a product that supports people facing loneliness, stress, or emotional fatigue, contributing positively to mental health and wellness through accessible technology.

What's next for Mendi

✅ 1. Real-Time Avatar Animation (Live Mode) Use real-time avatar animation that mimics user facial expressions during live sessions—deepening the sense of connection. ✅ 2. Emotional Journaling with Voice Notes Allow users to record and archive emotional voice logs. The system could analyze trends and offer reflective insights over time. ✅ 3. Multi-language and Cultural Sensitivity Support Expand language offerings with culturally nuanced responses and avatar styles, fostering inclusivity for diverse user backgrounds. ✅ 4. Companion Memory and Personalization Implement long-term memory so the AI recalls names, past conversations, preferences, and recurring emotional states. ✅ 5. Voice Cloning for Familiar Comfort Allow users to optionally use a familiar voice (e.g., a loved one or chosen voice clone) to create a sense of reassurance. ✅ 6. Integration with Wellness Apps Sync with mood trackers, calendars, or mindfulness platforms like Headspace, Fitbit, or Apple Health to personalize support suggestions. ✅ 7. Crisis Mode with Escalation Protocols Detect crisis language and offer predefined escalation pathways, such as calming routines or referrals to live mental health professionals.

Built With

Share this project:

Updates