Sonic Telegram

• 

  Sonic Telegram

Inspiration

As avid HIIT enthusiasts, we were inspired to create AI-generated workout music that adapts to each user's training needs. We wanted to build an app that motivates people with personalized soundtracks based on emotion, tempo, BPM and mashups of tunes. Having done HIIT ourselves, we're aware of copyright issues around workout playlists and saw an opportunity.

What it does

In 24 hours we hacked together a proof of concept app that takes in parameters like emotion and BPM, then outputs AI-generated music tailored to those workout inputs.

How we built it

We used Streamlit for quick prototyping of the web app UI. Python libraries like PyAudio, librosa, numpy and mido enabled rapid integration of customizable AI music generation.

Challenges we ran into

In just 24 hours, we didn't get to integrating full biometric tracking per user, but got the core functionality working for adaptive music based on various inputs.

Accomplishments we're proud of

We successfully built an end-to-end adaptive music app in just 24 hours including emotion, tempo, user set BPM, mashup, changing duration and a plan to integrate this as a Hugging Face model. We proved the concept of motivational, personalized AI-powered workout music.

What we learned

Through rapid prototyping, we learned how quickly we can build AI apps with modern tools like Streamlit and Python libraries. Iterating and team collaboration is key.

What's next for Sonic Telegram

We hope to refine this into a Hugging Face Model for workout music. We plan to integrate biometric data per user and continue improving the AI model quality before public deployment.

Please let me know if this summarizes well based on the details you outlined. I'm happy to expand or modify any part as needed!

Built With

• python
• soundcloudapi
• streamlitio