Open fquirin opened 6 years ago
hello florian, thanks for those hints :) it's a bit deprecated anyway, sorry..
thanks for the SEPIA-link, looks interessting and i will try it..
My hopes currently lie in Mozillas DeepSpeech
did you see mycroft.ai? they are working on integrate this. no clue, how it works, but maybe this helps you as well..
i will change that name soon, or maybe delete the complete repo, what you like. it's deprecated anyway, so i better fork SEPIA ;)
About the repo: it was nice to find it here, actually I forked your "fork" again to make kind of a backup ^_^. Feel free to do with it whatever you want ;-)
About mycroft.ai: I visit their page from time to time and read their blog article about DeepSpeech a while ago: https://mycroft.ai/blog/deepspeech-update/ - Seems they are putting quite some effort into it but are not really happy with the results yet. That was also kind of a damper for me because I had high hopes that there was finally something out there that could compete with Google, Microsoft or at least Amazon. Looks like they are still missing a lot of training data and tuning of the engine :-( In SEPIA I wanted to focus more on the user-experience than fighting with the open-source ASR, but it is something I'm really missing to complete the whole package. Also I thought it was kind'a cool how it was implemented in ILA :-) ... let's see how it evolves ^^.
Btw I saw somewhere on your page that you've worked on replacing google-maps with open-street-maps for the Mycroft project ... did that give you equally good results? I was thinking about doing the same thing for SEPIA and also thinking about open-weather-map as a replacement for DarkSky.
Hi Hersche,
I was just checking my 3 years old ILA project and realized that the download links were broken, probably due to my old provider changing it's ftp address. Since I was too lazy to dig through my old backups I just googled the file name and was very surprised to find that someone had "forked" the project on GitHub :-) Awesome!! And thank you for the nice description, but ... my name is Florian not Fabian :-p
I've been working since on a successor of ILA called SEPIA :-) It's trying to keep the spirit of ILA alive by implementing the same principle of teaching new commands (which eventually arrived in a similar way now in Siri on iOS 12 as well,... 3 years later ^^), but it is designed to work as a cloud-service running as your own private server on e.g. a Raspberry Pi. I somewhat gave up on Sphinx though and got stuck with platform-native ASR on Android, iOS and Chrome. My hopes currently lie in Mozillas DeepSpeech or Kalid to revive that part of the project.
SEPIA is close to it's first official release now. I'm still working on the "easy-installation" part (and GDPR policies O_o) but if you are interested you can have a look an leave me a note: https://github.com/SEPIA-Framework - GitHub repos https://sepia-framework.github.io/ - Homepage