MikeHG | # 851 | 2015-08-12, 18:15 | Report

Even doing this via a config file would be great as an interim measure. Fahrenheit is completely unintelligible to me, unfortunately (though I can still grok miles... we're an odd mix here in Britain).

Also, would it be possible to add a 'Cancel' keyword? I keep setting alarms for random times accidentally, being able to undo them would be awesome.

Great work so far though!

peterleinchen | # 852 | 2015-08-12, 18:56 | Report

Does Saera learn my speech pattern and pronounciation?
And is a 'Saera, what can you do?' implemented, like a help with all available commands?

taixzo | # 853 | 2015-08-12, 21:39 | Report

I'm working on a config file for the next release.

There is a 'Cancel' keyword but at the moment it only cancels navigation - I will add the ability to cancel other things.

Saera is currently running off a fixed voice model - the software I am using is almost certainly capable of rebuilding a custom model, but I do not know how to do that yet.

The latter is not implemented yet, but that is a good idea, and I will add it to the next release.

taixzo | # 854 | 2015-10-04, 18:51 | Report

New release uploaded to OpenRepos. This should fix errors for people who do not have Jolla Core apps installed, or do not have any accounts in them.

carepack | # 855 | 2015-10-05, 09:00 | Report

hi taixzo,

thx again. But for me and also the comments on openrepos this problem is still present:

Connecting to pyjulius server
ERROR: Error in loading model

Since you released the first public version for sfos no version worked for me. Hope it will be fixed in the next update.

Best wishes from my side

Greendts | # 856 | 2015-10-05, 09:39 | Report

I had similar issues for most releases, but another problem arose with my Jolla and I had to factory reset. After doing that, Saera works fine for me. Not a good work around, but it worked for me

arthurhkt | # 857 | 2015-10-05, 13:29 | Report

Hey Taixzo, Thank You for hardwork. However, unfortunately for me, after few reformat and installation, it's still not working for me. It's always the same problem for me, stuck at the "intializing speech recognition". I had install espeak and portaudio, but it never work for me.

V10lator | # 858 | 2015-10-06, 12:00 | Report

Same here. After removing (commenting out) line 138 from /usr/share/harbour-saera/qml/pages/pyjulius/core.py the app finally starts but now it looks like nothing is working. Also log tells
NameError: name 'saera2' is not defined
and other errors that seem to be connected to saera2 now.

taixzo | # 859 | 2015-12-14, 05:05 | Report

New release will be on its way in a few days. No fix yet for everyone with broken speech models, but I am improving the way Saera stores logs so I can get more useful error reports.
Also, I realized that, if every night you tell Saera "Wake me at eight thirty", you will end up after a few weeks with dozens of alarms set for 8:30. So now Saera cleans up these alarms after they go off (but only the ones she sets).

taixzo | # 860 | 2015-12-20, 16:05 | Report

Saera v2.2 is released to OpenRepos!

Changes:

• Saera cleans up after her own alarms
• Config file added (in /home/nemo/.config/saera/setting.json)
• Support for Internet sources for voice recognition added - currently only Wit.ai has been implemented. This is DISABLED BY DEFAULT to preserve user privacy; to enable it, set internet_voice to true in the config. It will only use internet voice if wifi is connected, as it uses a lot of data.
• Saera will now mute the audio of any currently-playing music while listening, to improve recognition accuracy.
• Help has been added - ask Saera "Help" or "What can I say" and she will give you a list of suggestions.
• Fixed navigation
• Fixed handling of contractions ("What's", "who's" etc.)