A handwriting Recognition System with text to speech. We used the already existing MNIST dataset for the numbers 0-9 and then also a free kaggle alphabet dataset
https://www.kaggle.com/sachinpatel21/az-handwritten-alphabets-in-csv-format
To train our model. we achieved a traning accuracy of 98.67%, while a validation accuracy of 80%. we then fed the model into a flask file which used gTTS to convert the predicted natural characters to speech.