Deep Learning is prevalent nowadays. As a crucial part of that, Natural Language Processing is also be focused. Inspired by studying Recurrent Neural Networks, we decided to use it to build a language model that can generate text with Shakespearean Style.
What it does
By feeding Shakespeare's work, the model will learn the way that Shakespeare writes. Then, once we apply a seed to the model, the model will automatically write the following parts.
How we built it
Keras library + Flask Web Framework + Web
Challenges we ran into
Do not have enough time to implement a more sophisticated architecture which has a better performance in text generation. We compromised the performance and build a word level-model rather than combining word level and character-level.
Accomplishments that we're proud of
Finished the project on time.
What we learned
Cutting-edge knowledge about NLP, Fast development skills
What's next for Shakespeare Or Fakespeare
- Improved the performance by using Gated LSTM.
- Improved the user experience.
- Used Google translate to read the text and record the voice. Then use transfer learning to modify the audio and makes the voice sounds like an opera.