This dialect detector predicts which dialect of Spanish a sentence is from. For the moment, it only distinguishes between Mexican Spanish and other varieties. I trained this model using text data from two sources (childes.talkbank.cmu and radioambulante.org).
To use the dialect detector:
- Launch the project in Binder (may take several minutes):
- Open the jupyter notebook 'dialect_predictor.ipyn'
- Run all cells in the "Intro" section.
- In the section "Type sentence, get dialect," type in a sentence and run the cell to get a prediction.
To see the code interactively:
- Launch and open the jupyter project in Binder
(may take several minutes):
- Open the jupyter notebook 'capstone.ipyn'
- You can either (i) run all cells (this includes the time-consuming process of scraping and cleaning the training data), or (ii) skip to the section "Data wrangling" > "Full dataset" > "Read in full dataset from disk" (this skips data scraping but still includes model fitting), or (iii) skip to the section "Feature engineering" > "Read in training data from disk"