- Improve the dataset
- More neutral sentences in data collection
- Larger dataset for better accuracy
- Annotate the dataset (done)
- Improve the User Feedback
- More analysis on personalized suggestions
- Detect content of the speech, remind users when casual words are used
- Improve the Model
- Use spectrograms to train the model
- Measure the model accuracy
- Improve UI/UX
- Compatible with every devices including mobiles and tablets
- Many different types of feedbacks depends on the users' politeness