AstroHealth

Inspiration

Astronauts have to operate in extremely stressful environments in space, which not only takes a toll on their mental health, but is also detrimental to the overall space mission. A common way to reduce stress and improve the mental state, is through journalling which provides them a vent and an emotional outlet.

What it does

AstorHealth generates music based on emotions experienced by the Astronauts, to elevate their mood.

How we built it

Using the text of the journal, the sentiment and emotions of the astronaut are detected. Using the sentiment a generative model is used to get the qualities of music which could help elevate the mood of the astronaut. Using the generated prompt, it is fed into a MusicGen model, which outputs the final music.

Challenges we ran into

Coming up with a way/input to detect emotions and understanding the mental state of the astronaut.

Accomplishments that we're proud of

An end-to-end workflow, starting from text data to generated music.

What we learned

A ton about the challenges faced by astronauts, and the difficult conditions they have to work in.

What's next for AstroHealth

Adding more modalities, and further fine-tuning to conditions in space. Currently, the model runs on cloud, but ideally it should be deployed on device, to allow for local execution.

Built With

• gpt3.5
• humeai
• meta-musicgen
• python