Inspiration
In the Philippines, there over 8 dialects and since it is customary to visit our relatives in the provinces. Unfortunately they speak Ilonggo! There are not much apps that translate from english to Ilonggo so i made one since Google Translate only has Cebuano-English
What it does
Currently it accepts a .wav file in english and translates it into Ilonggo text
How I built it
The model was made using a seq2seq model and a training set of 200+ english to ilonggo phrases. I used nodejs to accept a .wav file that uses Speech-to-text API in GCP and a jupyter notebook that act as server to do the translation with the model
Challenges I ran into
It was hard to acquire the training set as there were no existing ones. Another challenge is deploying the model as the code did not have a compatible model to deploy with Tensorflow-serving. So i improvised and created a microservice embedded in the jupyter notebook that was used in the REST API
Accomplishments that I'm proud of
I am proud that i was able to figure out a way to deploy the model which was coded using tensorflow checkpoints
What I learned
It was a challenge to deploy the ML models especially if the training method used was not compatible with Tensorflow-serving. It also require a lot of time to look for available datasets to train the model.
What's next for English to Ilonggo Translator
- enhance the input to accept realtime voice and do speech-to-text on the fly
- additional ilonggo word phrases

Log in or sign up for Devpost to join the conversation.