OpenUp

Inspiration

We built a hack that helps therapy patients keep track of their mental state in between therapy sessions

What it does

OpenUp lets patients record audio files following an emotional/traumatic event. Based on the transcription, OpenUp predicts the patient's emotional state and sends a summarised transcription of the audio file is sent to the therapist.

How we built it

We used Google Cloud's speech-to-text API and Co:here's sentiment analysis and text summarization models to build our project.

Challenges we ran into

Training the text summarization model was difficult because there was a lack of datasets focusing on mental health for text generation.

Accomplishments that we're proud of

The fact that we did this as a two-person team! Also, working with emotional classification models was really cool :)

What we learned

How to build, train, and fine-tune an NLP model!

What's next for OpenUp

Speech-to-text conversion and text summarization that works for multiple languages.

Built With

• api
• audio
• cohere
• data-science
• emotional-classification
• figma
• google-cloud
• google-web-speech-api
• machine-learning
• matlab
• mp3
• natural-language-processing
• numpy
• python
• pytorch
• text-summarization
• wav