homophones.csv is a curated list of homophones in the following format:

• each line has several words that have the same pronunciation
• each word is separated by commas (no additional formatting)
• only ASCII characters are valid. Offending lines can be found using GNU grep:

grep --color='auto' -P -n "[^[:ascii:]]" homophones.csv

Here are some sample lines:

bolder,boulder
bomb,balm,bombe

They are sort of in sorted order, but no guarantees.

Feel free to contribute by making a pull request and following the formatting above.