Inspiration
When my grandfather had a stroke, we quickly realized how inaccessible speech therapy can be. The high costs, long wait times, and limited availability made it nearly impossible to get him the help he needed. Millions face the same struggle, whether due to speech impediments, neurological conditions, or confidence issues. We wanted to create an affordable, accessible, and engaging solution for speech improvement—Talkify was born.
What it does
Talkify is an AI-powered speech therapy assistant designed to help users improve their speech through personalized lessons, real-time feedback, and speech tracking. It analyzes speech patterns, identifies areas for improvement, and tailors exercises based on user needs. Our AI chat assistant provides instant support, making consistent practice easier and more effective.
How we built it
We developed Talkify using a combination of AI-driven speech recognition with a combination OpenCV, PyAudio and MediaPipe, an LSTM model to train and analyze speech impediments from the MFCC features, used the features of a pretrained model using API calls to help generate lessons, which helps improve speech patterns. The frontend client was built with ReactJS and CSS, ensuring a smooth user experience, while the backend server leverages Flask for speech analysis and lesson generation.
Challenges we ran into
- Fine-tuning AI to accurately recognize various speech impediments.
- Ensuring real-time feedback was both precise and helpful.
- Designing an intuitive and engaging user interface for all age groups.
- Integrating accessibility features to make the platform inclusive.
- Integrating the speech model into the front end and the back end to ensure seamless access from the user end.
Accomplishments that we're proud of
- Successfully developing a working prototype with real-time speech analysis.
- Implementing personalized lesson plans based on AI-driven insights.
- Making speech therapy more accessible to people who need it the most.
What we learned
- The complexity of speech recognition and the LSTM model to train and analyze speech impediments for diverse speech patterns.
- The importance of user-friendly design in accessibility-focused applications.
- How AI can be leveraged to make a real-world impact in healthcare and education.
What's next for Talkify
We plan to expand Talkify by:
- Enhancing AI accuracy with more diverse speech data.
- Developing a mobile app for greater accessibility.
- Partnering with speech therapists to refine lesson plans.
- Adding gamification to make speech improvement more engaging. Our goal is to make Talkify a go-to tool for anyone looking to speak clearly, confidently, and naturally.
Log in or sign up for Devpost to join the conversation.