Inspiration
We were inspired to create SpeakSharp from our own experiences and challenges as international students learning a new language and communication skills. Existing solutions like pre-made lessons and AI chatbots were insufficient - we realized that the most effective way to improve is by learning from your own real-life conversations and contexts.
What it does
SpeakSharp is a digital language and communication coach that utilizes your personal daily conversations as learning opportunities. With your consent, it listens to your speech, filters out background noise, and provides multidimensional feedback on your communication style, sentence structure, vocabulary, grammar and more. You can conveniently review this personalized feedback during or after conversations to improve your skills.
How we built it
We built SpeakSharp using a tech stack including Python, Django, Postgres, and Docker on the backend, and Flutter and Dart on the frontend. At the core is a state-of-the-art voice diarization pipeline optimized for in-person conversations.
Challenges we ran into
Key challenges included developing the voice diarization to accurately separate speakers and filter background noise in real-world environments, and designing the feedback system to provide helpful insights without disrupting the flow of conversation. Balancing coursework with dedicating over 50 hours per week each to the project was also demanding.
Accomplishments that we're proud of
We're proud of launching a demo version and conducting a successful pilot study with 20 users who are meeting with us weekly to track progress and give feedback. The positive reception and insights gained from real users validates the potential of our approach. Building out an initial version of the complex voice pipeline and feedback system in just 1.5 months while balancing a heavy course load is also an accomplishment.
What we learned
We gained valuable insights from our user research, including the ineffectiveness of existing language learning products, the willingness of users to pay for a solution that truly meets their needs, and the power of using one's real-life contexts for learning. Technically, we pushed our skills in voice AI, natural language processing, and app development. We also learned the importance of user testing and iteration.
What's next for SpeakSharp.io
We plan to expand on our demo to implement key features like real-time feedback, calendar integration, and premium tiers. The next major step is conducting a larger beta program to gain more user insights and further validate product-market fit and monetization. Long-term, we aim to be the leading platform for personalized language and communication coaching. We also see SpeakSharp as the first step towards a full suite of intelligent agent assistants that our initial user base will adopt as they graduate and progress professionally.
Log in or sign up for Devpost to join the conversation.