Notices


Reply
Thread Tools
Posts: 31 | Thanked: 38 times | Joined on Dec 2010
#61
Originally Posted by Blaizzen View Post
wow very nice! I've got to try this. Thanks heaps for the idea and implementing it Just wondering though how do you call it from javascript? Since I've never played with servers or have much we experience at all (i've just been a bit lucky so far with this )

I've actually implemented the flash idea I said before but your idea is heaps better, especially since the maneuvers also contain data such as street names, hence we can eventually get spoken streets and directions


for people interested, this is what is possible and soon to be much better once the server is set up
http://www.youtube.com/watch?v=CC5vSv1dppc
....
Oh wow, didn't realise that you actually got the flash approach working - maybe the python server is not needed then. Anyhow, usually I use jquery to do the http calls, but a quick google for 'xmlhttprequest firefox' revealed:

https://developer.mozilla.org/en/using_xmlhttprequest

which would mean something like:

var req = new XMLHttpRequest();
req.open('GET', 'http://localhost:9999/turn_right', true);
req.send(null);

(to make an asynchronous request, e.g. the javascript continues and doesn't care about the meaningless output from the server)

Cheers,

Joerg
 

The Following 2 Users Say Thank You to jhb For This Useful Post:
Posts: 2,829 | Thanked: 1,459 times | Joined on Dec 2009 @ Finland
#62
Awesome work. I had to say that I would like to hear pre recorded files on this one. This robot sound reminds me of some pre historic windows 3.11 text to speech applications and most of time makes me wonder "Was that r2d?"

I think some community members have posted their voice files to this thread:
http://talk.maemo.org/showthread.php?t=53845

But this hacking what you are doing is that way too awesome Thank you for that.
__________________
TMO links: [iSpy] - [Power search] - [Most thanked] - [Cordia - Maemo5 UI on top MeeGo Core] - [CommunitySSU]
 

The Following 3 Users Say Thank You to slender For This Useful Post:
Posts: 52 | Thanked: 18 times | Joined on Mar 2010 @ Cordoba,Argentina
#63
Amazing job folks! Any progress with the offline navigation and search? Is this even possible?

Thanks a lot again!
 
Blaizzen's Avatar
Posts: 397 | Thanked: 802 times | Joined on Jan 2010 @ Sydney
#64
Originally Posted by slender View Post
Awesome work. I had to say that I would like to hear pre recorded files on this one. This robot sound reminds me of some pre historic windows 3.11 text to speech applications and most of time makes me wonder "Was that r2d?"
I gotta agree, I've finished implementing the ovi maps with full speech using espeak (including street names thanks to jhb, again thank you!) and I cannot understand much . The turn left turn right part is understandable, but I think it can't manage the street names very well

Originally Posted by esiravegna View Post
Amazing job folks! Any progress with the offline navigation and search? Is this even possible?
Offline navigation is already available due to the plugin. I personally haven't looked deep into it yet and I might not get around to it for a while, however there's heaps of talented programmers here now modding ovi, so who knows what will happen




I modified jhb's python server slightly to allow for full sentences from the javascript to be sent, hence enabling speech for street names too. I've uploaded the modified html file and the python file for people to test.

All thanks goes to jhb and CormacB for doing all the hard work

Edit: Should mention its only English for now, to change the language, search for "//show manuvours" in the html file to see the speech part and change the words accordingly.
Attached Files
File Type: zip html.zip (222.5 KB, 960 views)

Last edited by Blaizzen; 2011-01-13 at 12:08.
 

The Following 36 Users Say Thank You to Blaizzen For This Useful Post:
Posts: 31 | Thanked: 38 times | Joined on Dec 2010
#65
Originally Posted by Blaizzen View Post
I gotta agree, I've finished implementing the ovi maps with full speech using espeak (including street names thanks to jhb, again thank you!) and I cannot understand much . The turn left turn right part is understandable, but I think it can't manage the street names very well

Offline navigation is already available due to the plugin. I personally haven't looked deep into it yet and I might not get around to it for a while, however there's heaps of talented programmers here now modding ovi, so who knows what will happen

I modified jhb's python server slightly to allow for full sentences from the javascript to be sent, hence enabling speech for street names too. I've uploaded the modified html file and the python file for people to test.

All thanks goes to jhb and CormacB for doing all the hard work

Edit: Should mention its only English for now, to change the language, search for "//show manuvours" in the html file to see the speech part and change the words accordingly.
Blaizzen, I am so much looking forward to give this a litte test try (I need to go to the train station asap). Thanks a lot for your work on this!

My feeling is that we could maybe do a mixture of the prerecorded files and having r2d2 giving us a little riddle with the street names here and there :-). Regarding language - right now I would leave it with english, but later on work with command_identifiers in the index.html, and let the speech system figure things out from translation files (maybe using gettext or something).

Again, thanks a lot for your work and even starting the thread - without it I wouldn't have bought an n900 two weeks ago - and you prove me right to have done so :-)

Joerg
 

The Following 4 Users Say Thank You to jhb For This Useful Post:
Posts: 137 | Thanked: 150 times | Joined on Jan 2010
#66


I had been planning on putting together a python http espeak server at the weekend but you are way ahead of me.

What I would like to do:
1. An event for messages like "After 200 meters turn left"
2. Then when close "turn left"
3. A "2D" icon display for the next maneuver rather than cluttering up the map
4. Update the distance after each maneuver.

As you might have guessed the detection for when we pass a maneuver is not very clever: I just compare the straightline distance from the current position to the maneuver to a fixed value. I think this should work reasonably OK in most cases with a reasonable threshold, and figuring out how to measure the distance accurately along the road would be a lot more work.

It should be possible to use the maneuver getDescription() method with the translator object to generate localized messages.
 
Posts: 1,203 | Thanked: 3,027 times | Joined on Dec 2010
#67
Blaizzen:You've been playing Assassins Creed, I've been playing GT5 and messing with SDL 1.3. The soundmanager js library is something I was looking at before christmas, just wasn't sure what performance would be like.

With regard to pre-recorded voices, as Nokia Maploader can download maps to N900, can it also download the voice files. If so you've got translations for lots of languages ready to roll. Could then use surfer dude!

As I mentioned previously, there exists an image for a traffic button. As a simple implementation, do you think it would be able to download an overlay image from say Google and draw it over the map. Seem to remeber seing a version of Maps on Symbian using different colours (red, yellow, green) for route depicting traffic conditions. Could be good.

As several of you seem to have a grasp/more experience of javascript, is there any code that can be cleaned/removed/optimised? What about ordering the file so that desktop code is in one place so that it could easily be removed when copying to the device to make loading quicker.
 
kazuki's Avatar
Posts: 115 | Thanked: 18 times | Joined on Jan 2010
#68
this is so awesome, it just brought life back into ovi maps again. i actually find myself using it again.
 
Posts: 137 | Thanked: 150 times | Joined on Jan 2010
#69
Here is an updated GuidanceModel http://pastebin.com/gphsHEGN and ApplicationPlayer http://pastebin.com/8bpYVSds which does advance warnings (with next street name) and maneuvers using the python espeak server. It uses the localized messages so it should work in languages other than English with the appropriate espeak language. Firrewall issues prevent me uploading the whole file right now.

I reorganized the code a bit so that the GuidanceModel sends events rather than directly calling GUI related methods, which I think is a bit cleaner. The voice event handlers are just stuck into ApplicationPlayer which is not so clean. There isn't any icon display yet in this version.

The distance code is very rough so this will be easily confused to turns that are close to eachother and inaccurate GPS.
 

The Following 4 Users Say Thank You to CormacB For This Useful Post:
jd4200's Avatar
Posts: 451 | Thanked: 424 times | Joined on Apr 2010 @ England
#70
Maybe you could use natural voices from AT&T, they are very human like:
http://www2.research.att.com/~ttsweb/tts/demo.php

Though the restrictions would mean you couldn't distribute them, and people would have to obtain them themselves.
 

The Following User Says Thank You to jd4200 For This Useful Post:
Reply


 
Forum Jump


All times are GMT. The time now is 13:36.