Project
| # | Title | Team Members | TA | Documents | Sponsor |
|---|---|---|---|---|---|
| 41 | Emotional Intelligence Device |
Jonathan Fouk Matthew Palmer Vivian Tseng |
Jacob Bryan | design_document0.pdf design_document0.pdf final_paper0.docx final_paper0.pdf presentation0.pptx proposal0.pdf |
|
| Emotional Intelligence Members: Jonathan Fouk (fouk2), Vivian Tseng (vtseng2), Matt Palmer (mpalmer2) Experience: ECE417 (Multimedia Signals), ECE420 (Embedded DSP Lab), ECE414 (Biomedical Instrumentation) Problem: People these days spend more and more time in front of their computers rather than interacting with people. With lack of use, their communication skills deteoriate and their ability to identify their own emotions become harder. In the middle of a conversation somebody might think that they're feeling annoyed at the other person, but are they actually just stressed about something else? What if they were able to identify their emotion and therefore communicate better? Sometimes it is also useful for someone to find the root cause of their feelings. What if they were able to look back through their day, and see at what time they started to feel sad to see what caused their sadness? Autistic children also have trouble expressing emotion and communicating with people. What if there was a device that could help them identify what they're feeling? Solution: We have an idea to design a wearable device to identify your emotion and provide helpful cues for you to communicate better. We have researched 3 methods for emotion detection: analyzing voice data to determine emotion from the tone of voice, speech processing to group words together with different emotions, and measuring physiological signals to analyze stress/excitement levels. We propose to combine at least 2 of these methods together to detect emotion, then notify the user of their emotion discretely in the middle of a conversation and maybe also provide helpful words or phrases for them to better communicate their emotion. At the moment we are thinking of creating a bracelet that contains a microphone and a sensor to detect skin conductivity levels. LEDs or some other type of display can be embedded into the bracelet band to tell the user what emotion they're feeling and also offer words they can use to describe their emotion. This bracelet will also be linked to a smartphone app that keeps track of your emotional history. That way you can track your emotional health! Existing technology: Currently, most of the emotional intelligence technologies are being used for customer service or advertisements. There is some research being done into wearable emotion detection devices, however most of them only utilize body sensors and not voice. Most emotion detection advances these days are also utilizing facial detection instead of speech processing. Challenges: Main challenge is accuracy of emotion detection through speech data. Current research from Stanford University places distinct emotion detection to have around 40% error, while clustering the emotions together into common groups places the error at 10-20%. We are also not sure how well word association would work for determining emotions since we use most words in all types of contexts. So, in summary: Features: - Determine the user's emotion - Keeps a history of user's emotion so that the user can track the cause of the emotion - Displays emotion onto bracelet, and potentially some helpful phrases to help describe the emotion Classification methods: - Speech processing on voice data to determine emotion from tone of voice - Physiological signals through - Blood volume pulse - electromyography (EMG) - electrodermal activity (EDA) - skin temperature - respiration - Potentially word grouping to determine emotions, more research will need to be done first - Include previous emotions into classification, emotions that you have felt earlier in the day will have a higher percentage of showing up later in the day - consider top few results as possible solutions, maybe display two emotions with the more likely one being bigger Hardware Components: - Microphone - Body sensors - Display - Microprocessor |
|||||