kingofhenz/latintts

LatinTTS is a small Python project that turns a spreadsheet of Latin vocabulary words into a spoken audio study file. You give it a CSV file with two columns, the Latin word and its English definition, and for each entry it uses Google text to speech service to say the Latin word out loud in a classical pronunciation, followed by the English meaning. The results are saved as batches of fifty words at a time in separate audio files, which you then stitch together into one combined, compressed audio file you can listen to like a podcast while studying. Because the program has to make a separate request to a speech service for every single word, generating audio for a large vocabulary list takes a long time. The author reports that a ten thousand word list took around four hours to finish, so it is meant to be left running overnight rather than watched. There is a helpful link in the README pointing to a website with ready made Latin vocabulary lists to get started. There are two different scripts for combining the generated audio parts into one file. One only works on Linux and calls the ffmpeg command line tool directly, while the other does the same job in a way that works the same on Windows, macOS, and Linux, though either way you need ffmpeg installed on your computer. A couple of settings in the code let you adjust the playback speed of the Latin words and the length of the silent pause between the Latin word and its English translation, and between each vocabulary pair. The author is upfront that this is an early, unfinished project that has only been tested on Linux so far, and that a full rewrite is planned, with a beta release already available for anyone who wants the newer version. Bug reports are welcome as GitHub issues.