BodyBeats
Inspiration
In films and video games, the soundtrack is never fixed. During a tense horror scene, the music creeps in slowly, the rhythm tightens as danger approaches, and the whole room seems to hold its breath. The soundtrack responds to what is happening, making the experience more immersive.
But everyday music doesn't work that way. Your playlist keeps playing the same songs whether you're calm, stressed or full of energy. We wanted to bring that cinematic experience into everyday life by creating a soundtrack that responds to you. With BodyBeats, you are the scene.
We were also inspired by a familiar problem: finding the perfect song halfway through a run, or searching for something to give you that final burst of motivation. We wanted to make music more personal, responsive and effortless.
What it does
BodyBeats is an adaptive music system that responds to your body in real time. Using an embedded camera, a microcontroller and a simple breadboard circuit, it combines fingertip pulse sensing with facial-expression recognition to influence the music being played.
As your heart rate changes and your facial expression is recognised, BodyBeats adapts musical elements such as tempo, rhythm and mood. Whether you're warming up, pushing through an intense workout or slowing down to recover, your soundtrack evolves alongside you.
Instead of manually searching for the right playlist, BodyBeats aims to make your music respond automatically to your workout.
How we built it
We brought together embedded hardware, computer vision, signal processing and music generation to create BodyBeats.
- Hardware: We used an embedded camera and microcontroller alongside a breadboard circuit and fingertip pulse sensor to collect real-time data.
- Heart-rate monitoring: We processed pulse readings to estimate the user's heart rate and track changes in exercise intensity.
- Facial-expression recognition: We developed a machine-learning model to classify four visible facial expressions: happy, sad, angry and neutral.
- Data processing: We combined the sensor readings and expression predictions to create a changing representation of the user's current state.
- Adaptive music: We used this information to influence the music, changing musical characteristics to create a more responsive soundtrack.
Getting the hardware, software and music system to communicate reliably was a key part of the project.
Challenges we ran into
One of our biggest challenges was getting reliable real-time data from our hardware. Pulse readings can fluctuate, and camera-based facial-expression recognition is affected by lighting, positioning and movement.
We also had to balance responsiveness with stability. If the music changed every time a sensor reading fluctuated, the experience would become distracting rather than immersive. We therefore had to consider how to smooth incoming data and make musical changes feel intentional.
Another challenge was bringing together several different technologies, including embedded hardware, Python, machine learning and browser-based music generation, into one working system.
Accomplishments that we're proud of
We're proud of bringing together several different areas of technology to create something interactive and unusual.
We built a system that combines biometric sensing and facial-expression recognition to influence music, rather than simply displaying sensor readings. We're particularly proud of developing our own facial-expression classifier and integrating it into the wider BodyBeats concept.
Most importantly, we turned an idea inspired by cinematic soundtracks into a working prototype that explores how technology can make everyday experiences more immersive.
What we learned
We learned how challenging real-time systems can be, particularly when combining hardware, machine learning and software that all need to communicate with each other.
We gained experience with embedded development, computer vision, model training, signal processing and real-time data handling. We also learned that a machine-learning model's accuracy is only one part of the picture: its predictions need to be reliable enough to work in the environment where the product will actually be used.
Perhaps most importantly, we learned that good technology isn't just about collecting data. It's about turning that data into an experience that feels natural and meaningful.
What's next for BodyBeats
Now that we've integrated the Spotify API, our next goal is to make BodyBeats something people can genuinely use while exercising. We want to explore how to make the system accessible and convenient during a workout, so users can experience adaptive music without having to interact with their device.
We'll also focus on optimising the system: improving the accuracy and reliability of our heart-rate monitoring and facial-expression recognition, reducing processing delays, and making transitions between musical styles feel smoother and more natural.
Another priority is refining how the different inputs influence the music, ensuring that changes feel meaningful rather than reacting to every small fluctuation in sensor data. We'd also like to personalise the experience, giving users more control over how their physical activity and facial expressions shape their soundtrack.
Ultimately, we want to take BodyBeats from a hackathon prototype to a practical exercise companion that makes every workout feel like its own cinematic experience.
