|
|
|
|
|
by tekacs
4775 days ago
|
|
> Very personalised information wouldn't work. I have a built and working version of something not far off what the OP describes, only for myself (haven't had time to open source it or add a config interface) but that uses snippets of my own voice (for reminders) as well as Twilio's text-to-speech API to generate these messages (partially due to its providence as an alarm - TTS + avconv + e-mail would work for those who didn't want that). I've had to custom build the generators (templated text + scraping code using utilities) for each data source, but it's not much of a chore and indeed a one-time task. I write these by hand mainly because I think it's valuable to have audible cues (provided by the surrounding template) when hearing particular information and code + template makes conciseness easy, too (cf. IFTTT). I've thought in passing about the fact that people might appreciate its use, but I've felt inclined towards open source rather than subscription, at least in part because I haven't necessarily the time to make it configurable for the non-technical. I suppose throwing humans into the mix, as you suggest, would help with that. :) |
|