Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

Something is horribly broken. I don't know whether it's in pyotherside or just something wrong with my install; but certain lines of Python code are causing segmentation faults, especially anything that has to do with the Internet or Unicode, and I'm rather at a loss as to what to do about it, because as far as I can tell nobody else on the entire Internet has ever had a segfault with PyOtherSide. Development os Saera has been suspended until I can work this out.

Edit: thanks to thp for helping me with this one, the issue was that the emulator ships with old packages of PyOtherSide. Everything is good again! Development has been resumed!

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

And ported to Harmattan. After using PyOtherSide, PySide feels very sluggish; I think I may redo it with an old build of PyOtherSide from when it supported Qt4 instead.

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

Well done! Can't wait to give it a try.

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

still cant install this on my N9
would some one reupload the file dependence..?
plese. i cant donwloadbthe resource from here

How Can I install/get PyGTK2?
Iv look for it more than a month.
Nothing founf

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

should be possible to use mbrola for saera?

voices with mbrola+espeak are less robotic

this is a small guide to use mbrola arm version on the N900 for Navit turn-by-turn voice instructions

Download mbrola arm version from this link:

http://www.tcts.fpms.ac.be/synthesis...nux/mbrola.rar

i downloaded on my pc, extracted, and transfered the mbrola binary file via ssh to N900 in /usr/bin/ folder. (maybe is possible to save some rootfs space putting the mbrola binary in /opt and then create a symlink from /usr/bin to the file)

for voices i found to ways:

you can download voices from http://www.tcts.fpms.ac.be/synthesis...brcopybin.html

download it4 package, extract, cause you need only the it4 file. then put the it4 file in a folder in MyDocs.

mbrola will search voice files in /usr/share/mbrola/voices/ so you need to create a symlink /usr/share/mbrola/voices/it4 linking to the file in MyDocs

another way i want to test is to use some mbrola voice files already present in espeak folder /opt/espeak/data/voices/mb/ (so no need to download a voice file)

to let mbrola work with navit just edit speech.xml file like this:

<speech type="cmdline" data="espeak -v mb-it4 -s 150 -a 150 -p 50 '%s'"/>

and should work!

-s sets the speed
-a sets the volume(av. 100)
-p sets the pitch(av. 50)

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

Not sure if mentioned before, but would Jasper (http://jasperproject.github.io/) be of any use here.

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

.. any news?

Re: [DEVEL] Saera: Siri clone for Maemo5 and Harmattan
 

So it's just pocketsphinx running all the time - isn't Saera also using pocketsphinx already...? ;) But indeed maybe looking at the code and configuration of Jasper could be of help (how to solve some stuff, maybe also some dictionaries or phonetic models could be used, maybe some interesting ideas could be borrowed and reimplemented...).

edit: I think it is also worth noting that Ubuntu since 13.04 or 13.10 ship with pocketspinx by default, as they want to implement voice control in their Unity interface - maybe they have some nice models and dictionaries of phonetics in different languages which can be reused here.