SonicLearn AI

SonicLearn AI transforms speech into structured learning materials using ElevenLabs and DeepSeek, helping students and professionals learn faster through automated summary, glossary and references.

Comment 1

Inspiration

The idea for SonicLearn AI was sparked by a real accessibility challenge. During the COVID-19 pandemic, our teammate Nishit, who is deaf, could no longer lip-read lecturers due to mandatory mask-wearing. Unable to follow verbal instructions, he was forced to pause his education. This experience revealed how easily traditional learning environments can unintentionally exclude students. We came together to change that, not just for Nishit, but for anyone who struggles to keep up with audio-based instruction with an affordable solution.

What it does

SonicLearn AI is an audio learning assistant that uses ElevenLabs Scribe speech-to-text API and natural language processing API to turn spoken content, like lectures or webinars, into structured, digestible learning materials. It auto-generates transcripts and key learning insights to support inclusive learning for students with hearing challenges, English as a second-language, or anyone who would like an AI-powered learning tool.

How we built it

We built the frontend using Bolt with no code, integrating ElevenLabs Scribe speech-to-text API for transcription and natural language processing DeepSeek API (chosen for its accuracy and affordability) to interpret and generate subject-specific structured learning materials. The final app was deployed on Netlify for accessible, shareable hosting.

Challenges we ran into

  • Ensuring clean, high-quality AI-generated subject-specific summaries from messy or unstructured speech required multiple rounds of prompt engineering and output evaluation.
  • Technical limitation with ElevenLabs Scribe speech-to-text API, which is unable to transcribe live, only with recorded audio.

Accomplishments that we're proud of

  • Delivering a working prototype that automatically transcribes and summarises spoken content with no code.
  • Creating a solution that directly addresses a lived accessibility barrier.
  • Be cost-conscious, choosing DeepSeek for value for money, instead of more expensive LLMs such as GPT or Gemini who do not deliver enhanced performance despite additional costs.

What we learned

  • Designing for inclusivity from the outset, not as an afterthought.
  • Iterating on AI output to ensure clarity, brevity, and usability for learners.
  • Navigating unfamiliar UI ecosystems and adapting quickly when things (inevitably) broke.

What's next for SonicLearn AI

We plan to expand SonicLearn AI by:

  • Integrating with learning management systems to streamline academic workflows.
  • Working with educators and students to test real-world usability and gather feedback.
  • Allow live transcription once the ElevenLabs scribe live functionality is launched (which is advertised as "soon".
  • Deploy it as a chrome extension to make it even more accessible.

Our mission remains clear: break down barriers and empower every learner, whether in a lecture hall or learning from home, to access, retain, and apply knowledge more effectively.

Built With

Share this project:

Updates