Notices
Page 39 of 72 « First 293738 39 404149 Last »
Reply
Thread Tools
ruff is offline ruff
2016-05-24 , 19:03
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#381
while playing with this response action realized we don't have any voice support whatsoever, implementation for endpoint 0x2af8 is completely missing
 
abranson is offline abranson
2016-05-24 , 19:14
Posts: 287 | Thanked: 862 times | Joined on Dec 2015
#382
Originally Posted by ruff View Post
stockpiled quite a big bunch of commits, probably a good time to merge. Posted a beta rpm.
Atm - no (visible) new functionality - except everything timeline related is moved to timeline engine - notifications and calendar.
As a side effect - some colors are changed, some new actions are added, devcon supports pin-insert, there's dbus call to insert the timeline pin, notifications are pooled and re-delivered on connection...
Don't be confused by Canned Response action - it's part of notifcation now however platform part is not yet implemented, so it does nothing except notifying platform part to do the reply.
Next steps - platform part and websync.
That's looking really nice, huge drop in code complexity here. A merge as soon as you're ready would be good - I don't think we should release it until the websync is there as users won't see any benefit (though beta is essential). I'd like to help with the plumbing though. I reckon this will be a good time to drop 1.0.

It'll need pebble.com login too for the websync, right?

Originally Posted by ruff View Post
while playing with this response action realized we don't have any voice support whatsoever, implementation for endpoint 0x2af8 is completely missing
One thing at a time! :P
 

The Following 3 Users Say Thank You to abranson For This Useful Post:
ruff is offline ruff
2016-05-24 , 19:38
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#383
Originally Posted by abranson View Post
I don't think we should release it until the websync is there as users won't see any benefit (though beta is essential). I'd like to help with the plumbing though. I reckon this will be a good time to drop 1.0.

It'll need pebble.com login too for the websync, right?
Sure, it's just to reduce review volume later on one side, and get rid of an urge to fix something backdate on the other. While nothing wrong with that per se, it also affects reviewability of the change set.

Yes, it would need oauth token from pebble.com.
 

The Following 2 Users Say Thank You to ruff For This Useful Post:
ruff is offline ruff
2016-05-24 , 19:41
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#384
Originally Posted by abranson View Post
One thing at a time! :P
Absolutely, just to mark it somewhere till i remember that. Realized then there's better place to leave such reminders - issues board
 

The Following User Says Thank You to ruff For This Useful Post:
ruff is offline ruff
2016-05-24 , 20:19
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#385
Originally Posted by Bundyo View Post
Here we go again

https://www.kickstarter.com/projects...y-new-3g-ultra
crap, attempted to pledge for core - refuses my card. to hell with it.
 

The Following User Says Thank You to ruff For This Useful Post:
taixzo is offline taixzo
2016-05-24 , 20:30
Posts: 951 | Thanked: 3,409 times | Joined on Apr 2012
#386
Is there any way I can hook Saera into the voice things?
__________________
Saera: A Siri clone for the N900, N9(50) and Jolla

Follow me on Twitter: twitter.com/taixzo
 

The Following 4 Users Say Thank You to taixzo For This Useful Post:
ruff is offline ruff
2016-05-24 , 21:59
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#387
Originally Posted by taixzo View Post
Is there any way I can hook Saera into the voice things?
I don't yet understand fully how it works, but since there's word AudioCodec (speex) I assume that certain phone app should do actual recognition and submit results. And I don't think we have a big offer of speech recognition services on jolla
Voice session setup request looks like this:
Code:
VoiceControlCommand
CMD: 01
 SessionSetup
FLG: 00000000
data
 SessionSetupCommand
  Type: 01
   Dictation
  SesID: 0800
  attributes
   CNT: 01
   dict
    Attr
     AID: 01
     LEN: 1d00
      29 (LE)
     Attribute
      SpeexEncoderInfo
       VerStrn: 312e327263310000000000000000000000000000
        "1.2rc1"
       SmplRte: 803e0000
        16000 (LE)
       BitRate: 0032
        12800 (LE)
       BtStrVr: 04
       FrmSize: 4001
        320 (LE)
Last edited by ruff; 2016-05-24 at 22:45.
 

The Following 3 Users Say Thank You to ruff For This Useful Post:
abranson is offline abranson
2016-05-25 , 06:46
Posts: 287 | Thanked: 862 times | Joined on Dec 2015
#388
Saera is about the only option we have, because I think we can agree that Sailfish and cloud-based speech recognition are not morally compatible. Maybe I should pledge for a Pebble 2 so I can at least test it
 
ruff is offline ruff
2016-05-25 , 06:54
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#389
Originally Posted by abranson View Post
Ofono was the only way I could find to send SMSes. I hadn't found any way of sending IMs yet.
Another way would be using telepathy.
This is the one I've been trying to explore and think should be more appropriate as it allows:

a) step up one abstraction above phone implementation
b) use common interface for SMS and IM messages
c) it automatically adds messages to the comhistoryd - visible in messaging centre.

There was an issue though that only one consumer could be attached to Tp::account as it uses static member internally. However it seems (partially) this was fixed.
nemo-qml-plugin-messages is an example how it works.
Here's an example - https://talk.maemo.org/showthread.php?t=95876
You can get directly to the channel for the phone as it is hardcoded mostly (ring/tel/ring), but IM accounts should be properly discovered based on notification data (from that base64 encoded VariantMap in the x-nemo-remote-action-default) using appropriate telepathy management interface.
 

The Following 2 Users Say Thank You to ruff For This Useful Post:
ruff is offline ruff
2016-05-25 , 10:42
Posts: 207 | Thanked: 482 times | Joined on Mar 2016
#390
Originally Posted by abranson View Post
Saera is about the only option we have, because I think we can agree that Sailfish and cloud-based speech recognition are not morally compatible.
That was the essence of my previous message

So taixzo - while we're all here - can you briefly introduce us into interaction with your product? DBus calls? API?
I.e. can we just dump raw speex to /tmp/blah and signal Saera to perform recognition of the dump, then catch another signal in return and forward results to pebble?
 

The Following 2 Users Say Thank You to ruff For This Useful Post:
Reply
Page 39 of 72 « First 293738 39 404149 Last »

Tags
pebble, smartwatch

« Previous Thread | Next Thread »

 
Forum Jump


All times are GMT. The time now is 21:27.

Contact Us - Home - Archive - Top