Trained 4 models, male/female positive/negative emotion classification models and male/female multi-emotion classification models multi-emotion classification models perform poorly, so will be using binary classification instead Combined text and tone analysis Moved nets to cloud (but still need to work on getting it hooked up with the web app) Web app Read more…
This week I focused on completing integration and preparing for the final presentation. Last week I was struggling with how to divide the journal entries into sentences. Google Speech Recognition works best on 5 second snippets of audio. Additionally, the library is not able to recognize sentence boundaries. My solution Read more…
Redeemed our cloud credits Planned out integration Work on: Actually integrating all the components Making video Sharing code
Training the voice tone analyzer Keep working on: Integrating the team’s parts together! Put everything on Github to share with team members Put everything on cloud
I’ve reached 60% overall accuracy for the 7 categories (6 emotions + neutral). I’ve started to work on integration. I’m able to read speech from the computer mic and analyze it using python’s SpeechRecognition library. Right now, the program listens to a user until it hears a second of silence, Read more…
This week our group continued to work on our individual modules. Patrick trained around 60000 images for his facial emotion recognition and will add a secondary SVM for detecting emotions that are harder to tell apart. Vinay finished his first iteration of training for textual sentiment analysis. Yoojin added video Read more…
Additions to webapp, video streaming Reevaluated problem statement and goal through ethics assignment More tests with example voice tone analyzer using CNNs To Do: Journal recording storage on cloud Fix display bug for videos More Voice Tone Training
This week I finished the Google speech to text pipeline and a test script to demo my best network. I’m working on increasing my accuracy past 65%. I’m expecting to hit 70% accuracy next week. Additionally, I will work on creating a better dataset to train my final network on. Read more…
Changed the program I’m using to get pitch and other features; no longer using Praat because it takes too long and is inefficient; using the python Librosa library instead (documentation here https://librosa.github.io/librosa/) Testing out example voice tone analyzer using CNNs; one project claims to have reached 70% accuracy Next week: Prepare Read more…
This week I trained finished training a few networks. The highest accuracy I achieved was 53% which is less than ideal. I trained on 40,000 tweets scraped from twitter and did some minimal preprocessing. Going forward, I want to implement the Google speech to text pipeline and train the network Read more…