Waking up on time is a struggle many of us face, especially as heavy sleepers. We recognize the value of a morning routine, but conventional alarms often fall short. We wanted to create a solution tailored to this widespread issue, particularly among students. Studies also highlight that engaging in mentally stimulating activities or conversations can significantly enhance alertness, making this an essential part of our concept.
The app requires users to set an alarm and select a conversation topic. Once the alarm goes off, the user must engage in a voice conversation with Cohere’s LLM. The conversation continues until the LLM determines that the user is fully awake. Additionally, the LLM integrates with the user’s personal calendar, allowing it to answer questions about their schedule for the day.
We created a voice interface using a speech-to-text system to capture the user's input, which is then sent to the LLM. The LLM’s responses are delivered back through text-to-speech, creating a natural conversation experience. Data is streamed in real-time through our backend, ensuring low-latency responses on the frontend. For calendar integration, we utilized Cohere's Retrieval-Augmented Generation (RAG) feature, allowing the LLM to access and respond to schedule-related queries only when necessary.
The primary challenge was integrating all the different components into a cohesive system. Ensuring smooth communication between text-to-speech and speech-to-text systems, and managing real-time data streaming, required careful coordination.
We’re proud to have completed this project as beginner developers. Successfully integrating advanced technologies like speech-to-text and LLMs into a functional app was a major milestone for us.
We gained valuable experience in connecting frontend and backend systems, mobile app development, and working with asynchronous functions and Promises.
Our next steps include enhancing personalization features, improving wake-up accuracy, and exploring further integration with smart home devices.
Built With
- cohere
- elevenlabs
- javascript
- node.js
- python
- typescript
Log in or sign up for Devpost to join the conversation.