Interesting, it seems like #chinese #xiaomi #cameras are #censoring some #pictures https://old.reddit.com/r/Xiaomi/comments.. Apparently it could be a #secret #encoding in the #german #election #document causing this - likely one that isn't widely known about yet. I think #hackers should definitely #investigate this document more closely!
@barray on Fri Jul 30 16:32:43 UTC 2021 said: &eStill have no real answer to the #background #music for the #coffeespace #texttospeech #audio ... I thought about perhaps using some #procedural music, but it can be not so great to listen to, especially with #lowquality audio #encoding parameters. Looking at #ffmpeg this is currently #twochannel (didn't make a difference), 22050 sampling frequency and 40k #bitrate - derived through experimentation. There might not really be an acceptable solution to this problem...When I created the new #background #music for the #coffeespace #texttospeech #audio that reads articles, I never thought that I would end up writing an article longer than 45 minutes of reading. I would suggest that the current article is about half way complete from where it will be - that's really quite considerable. This is getting into the #book territory - some books are not this long. Pretty impressive if I don't say so myself.
Interesting #audio #encoding concept getting barely #audible #voice down to 450 bits/second: https://www.rowetel.com/?p=6212 The proposal is that this could potentially be used for #lora #realtime #communication over very #lowbandwidth channels.