Inspiration
Building a playlist and choosing the right type of music, specially to study, is time consuming. Most times we end up wasting precious studying time distracted by our music. Some of us just give up and study in silence. But music is a powerful tool! It should enable our concentration, not hinder it. It could also take us to an excited mood, a calm mood, help us recover from illness. Our brain outputs more information than we actually recognize and "use". The possibilities of our brain knowing what it needs before we consciously know it are endless.
What it does
Say you are studying for a test and have music in the background, Mōōd is there to chose a song that will improve your concentration, based on your brain activity measured through the Muse headband.
Through Mōōd , you set up a user account and indicate the Mōōd you need: studying, running, meditation… you name it! Each Mōōd has a default list of songs linked to this particular Mōōd , but fear not, you can add your own songs. You put on the Muse headband and hit play and let your brain waves do the choosing for you. Mōōd will use the feedback from your brainwaves and switch to songs scientifically chosen to improve your Mōōd. If a song gets negative feedback from your brain, Mōōd will transition to something more appropriate. If your brain likes what it’s hearing, then Mōōd will do its best to keeping your brain happy. Now, what works for you won’t work for everyone, so we rely on machine learning to start getting to know your brain’s likes, and dislikes, quicker.
How we built it
The muse headband outputs information as an OSC (Open Sound Control) file. We've created a script that listens to the server launched on one of our machines in order to extract necessary information. We grab the information related to brain signals, more specifically, the beta wave absolute values for each of the four sensors located on the muse headband as a CSV file. We then process the information at a rate of 0.033Hz in order to check the state of the user. If the value calculated is within a certain range, we do nothing. Otherwise, if the value calculated exceeds our calculated threshold, we change the song currently being played and analyze the user’s mood after 30 seconds to check if progress is made. If progress is made and the user’s state is back within the safe range, we keep the song, otherwise we change again until we find the user’s equilibrium state.
Challenges we ran into
Getting data from the MUSE. Using the Spotify API to customize what the user will listen to.
Accomplishments that we're proud of
We're proud of getting the data of the MUSE Headband! We're also proud of joining all these pieces of code together in such a little amount time.
What we learned
We learned how to work with servers. React, Java script. Accessing API. And of course, you meet interesting people in a Hackathon.
What's next for mood
It would be nice to get hardware better suited for this task, a better wearable. For now, on our end, we need to perfect the bands for which each type of music is assigned. This would require research on what values we are aiming for, for each particular mood.
Finally, instead of manually analyzing a user’s state we would have enough data to use Machine Learning in order to train our algorithm versus always calculating the user’s mood from scratch. This is where IBM Watson’s Classification type products would come in handy where we can feed it data when a user is in a study mood and when a user is deviating from that in order to decide what song to play next. In principle, once a user wears the headband for some time, the machine has enough information to predict a tailored playlist, so the headband is no longer necessary.
Log in or sign up for Devpost to join the conversation.