myVoice Open Beta (1 Viewer)

[Marcusb]

hey there
I think this was talked about a while ago, but one feature that would be great is a kind of debug mode in the configuration section where you could test the speech response there.
Maybe have an option to turn on speech control and have the PC type what it hears?
This would be excellent for the people with issues to determine where the issue was (IE with recognition or with the speech engine itself).
It would also be useful to find out what the computer thinks you are saying. You could use the computer's spelling in the options rather than the correctly spelled versions.

Anyway, that's probably my number one featuer request at the moment.
No idea if it's feasible or not.
I know there is the debug mode where the computer tells you what it hears, but I think this way would make it easier. Then again, maybe a plugin can't be activated in this way in the configuration app, then maybe it could be implemented as an option in the voice window section?

 

[booyakasha]

Marcusb: I like the idea Marcus. I did some preliminary testing in the voice window section, and here's an idea for implementation: In the voice window section there's a dictionary button that you can drill down to the various phrases it's supposed to recognize. When the user clicks on an item, the computer speaks the phrase followed by a "now you try" response. The user speaks the phrase into the mic, and the computer speaks back what it thought it heard.
For example, if the user drilled down to Dictionary->General Actions and clicks on "Play Music", the TTS engine says "Play Music" followed by "Now you try". After you have spoken, it responds with "You recorded the following message" followed by what it thinks it heard.

I'm also looking at if you can add the users specific pronounciation of the phrase to the engine, but I'm unsure of this at this point.

 

[Marcusb]

That sounds excellent

Do you also know if the engine will work in configuration too?
If so, it may be easier to have a tab for testing in the plug in page, to save having to open MP each time you want to trouble shoot why word may not be recognised.

I do like the idea of my Alice interactively helping me set her up

 

[booyakasha]

The engine will work in the configuration. Currently I just don't really start it as there wasn't much point as I didn't see a need for voice control of the config.
I'm successful in adding a "Train" item to the context menu of the various phrases. When you click it, it will say the phrase, pop up a dialog, and allow you to record your pronounciation of the phrase. Below are a couple of screen shots. The first shows the "Train" option, and the second shows the window that would appear after clicking it (and after the TTS of the phrase).

I'm actually thinking this is the better way at this point. At the end of the day, we don't care what the engine recognizes as a phrase, just that it does. This would allow a sort of override on those problem phrases. This wouldn't really work too well for any of the music items, etc, as they're variable, but it may help the engine better recognize various phrases as it will better understand the way you speak.

 

[Yodar]

Hey, some isseu I stumbled on to.

In some windows, like my pictures when you give the command "move down/up" its also selects the item. So, in the my emulator window the game starts right away after the move command is given. This can be a bit annoying sometimes.

 

[SwingSultan]

hi,

yet again i have two minor problems. when i wake my media center from the s4 standby (hibernate) it doesnt recognize voice commands. i always have to restart mp in order to make it recognize my pretty voice. the other thing is not really a problem of the plug in. i would like to configure a command that maximizes mp when minimized to tray. i tried several ids but couldnt get it right. my goal was to emulate the green home button on the ms remote.

i also have a suggestion which could give the voice plug in a nice semi-intelligent star trek computer feeling. it would be great to configure alternative commands for certain functions for example that the comuter recognizes both "volume up" and "increase volume". this should at least with general functions like volume control give the voice interface an intuitive touch, if alternates like increase volume and volume up are included in the standard config. i got the idea while i showed the vioce interface to two of my friends who just tried talking to the computer before i could give them any hints aubout the commands. after saying "computer" in a star trek like fashion they initially tried to feed the pc with intelligent guesses like "increase volume" and so on. the alternate phrases could also be combined with the training function mentioned above.

and yert again: dude you kick ass!thanks for making a great plug in, continiously delveloping it and being our on demand faq/manual.

 

[booyakasha]

Yodar: I couldn't reproduce the problem you're describing. If anybody else is seeing this, could you please post. I'll put some debug code in the next release to see what's happening.

SwingSultan: Does your myVoice.log have any lines like "Stopping process for power suspend" or "resuming process for power resume" after the system is suspending or resuming? As for the minimize/maximize thing, I imagine those are API calls which wouldn't be handled normally. I'll look at that when I get a chance.

The last point about adding phrases: You can already do this. In the plugin config, go to the phrase you want to duplicate, right click it, and make note of the settings. In the case of "volume up", you have an ID of 103, and an option of NUMBERS. Go to the top of the section "General Actions", right click it, and select "Add". Type in the new phrase "Increase Volume", press enter, and then right click and enter the values noted earlier. Then press apply and you'll be good to go. You can do this anywhere at the phrase level to increase the vocabulary. I only released a general structure for reference, so people can add the vocabulary closer to how they speak. I have a friend who insists on saying things like "pump up da volume" and expect the volume to increase...lol.

oh, and thanks for the props.


Edit:

I've added the training feature, and it appears to be working, but I'd like some more feedback on it before I distribute it for wide release. If you have any trouble navigating, the training feature should help. If it's for songs, etc, it doesn't work as those aren't loaded in the config screen. I may be able to add that feature if there's a demand.

New Features:
-My Movies database integration: You can speak a movie by Title/Genre/Actor
-Music Album
-Training: Ability to train a phrase that the recognition system has problems understanding

Updated:
- TTS Debug
- Detail logging

Make sure to backup your equivalent files before copying over so you don't lose your setup

 

[Yodar]

whiee, another update. Ive been a bit bussy lately so wont test the RC1 out

Ill also see if I can reproduce my above described problem, It doesnt always happens.

Hope the training ability has any succes getting this plugin to work for my sister/mother. Since the computer responds to every guy and does not respond to any girl so far. Could be because our english pronucation isnt great. Yes, my sister had trained alot with her own profile, with no luck yet.

 

[ariejakkes]

Thankssssssssss for this great pluginn, i use MP 0.2.3.0 rc 3 and test myvoice 206 rc1 it works fine!!!! music artists works fine only when i say musicsongs...... the computer voice say it again but nothing happend ,its strainge but i look what i can do about it . In myvoice .xml it starts wih <?xml version="1.0"?> and i think it must start with <?xml version="1.0" encoding="utf-8"?> but when i change it it change back in de first one. anyone know what i can do about it( sorry my englisch is verry bad)
Many Thanks for making such a good plugin!!!!!!!!

AHH , me again , with an other question: when i say what time is it , computer say time is.... but not THE Time .do you Booyakasaha,do you know how i can fix that???/

 

[booyakasha]

Here's Release Candidate 2. Included are a bunch of bug fixes, as there were problems as ariejakkes described.
One known issue is if you are in full screen video and you want to play music, it won't work.

Also, if you notice things aren't working, please let me know, either through this forum or PM. I'm not really watching any TV or movies or listening to music from my HTPC at the moment, so it's very difficult for me to test everything.