Support iPod Voiceover
Some iPods (at least the iPod Shuffle gen 4) have a voiceover button: when pressing it, the device will spell the artist's name and the song's title, or the name of the playlists. It works by creating small audio files with TTS and storing them in iPod_Control/Speakable on the device.
This script supports creating and storing the voiceovers using several FLOSS TTS options: https://github.com/nims11/IPod-Shuffle-4g. Unfortunately, I tried eSpeak and it sounds horrible, the voice is extremely robotic and barely understandable. The scrip also supports picotts and its pico2wave binary, which if I believe this video, sounds great! Here are other example: https://www.youtube.com/watch?v=0sAhnxg5AD8, https://www.youtube.com/watch?v=T0pn1yEumTs. Unfortunately I didn't manage to make picotts work by building it from https://github.com/naggety/picotts, but it is packaged for Arch here.
It'd be really nice to have Rhythmbox produce and store these voiceovers when exporting files to iPods supporting it.