Livrot Mic Command - Preview Release
-
Thanks to Wingedsky for pointing out another bug.
Synonym display and variables display can now be left on while editing. All resulting changes are directly shown in the commandlist.
New Version 1.141008 is ready for you. It shall be the last beta version of LMC.Livrot
PS:
for those who have missed it - speech recognition of LMC is working excellent again since V1.140927
-
Somehow the speech settings (babble timeout, … ) got corrupted in the last profile.
New Falcon BMS example profile uploaded (V2.1). Please check your ‘settings’ / ‘speech’ slider positions and put them on the black marks if you have downloaded version 2.0 and had the ‘settings’/‘main’/‘use profile settings’ checkbox checked.I have finally given the ‘hotkey device manager’ a rewrite and it will pop up only if you load a profile made by someone else which actualy contains low level hotkeys recorded with devices unknown to your system.
Before it popped up even if no such hotkey existed in the profile.
Now unknown devices are removed from the hotkey device list upon first saving.This will come to you in the next version together with a straightforward tooltip about the ‘hotkey device manager’.
Thanks for your support, gents!
-
thanks for your effort Livrot,
each version is going better and better:), Bravo
As I am particularly interested with the response possibility of your soft, I noticed that you haven’t yet implemented the possibility to enable numerous responses with only one box. It is really annoying to click more than 100 time the same box, one by command…
But the main concern is once responses are enabled, profile saved with profile settings, when I reload the profile, every responses are off…:(
A second point : could you precise if there is a limit with the number of variables in a phrase? As I understand things, the variable is in fact something that you can change easily, but once defined, there is no particular search generated by the variable. Thus, no limit… which is different than synonym. is this correct ?Windblow
-
I will fix the enable all checkbox today, also the save / load problem that you described. Thanks again.
Variables are just a group of options. The user decides which option will be used when replacing the placeholder. Only the selected replacement option will be used.
So before every mission make sure that _tower, _alternate, _awacs, _callsign, … are the same as in your briefing.
Variables where created to permit real world coms while keeping the commandlist (as well as the resulting speech grammar) very compact.
All synoynms of one group are available for speech recognition, while only the selected replacement option for placeholders will be used.If any synonym of a synonym group is found inside a phrase all other synoynms will create new phrases replacing the found synonym.
If you want only the topmost synonym in a group to be searched for and replaced in phrases check the topmost checkbox. You would like to do so
when the replacements contain other synonyms which in turn would trigger unwanted new synonym production.There is no risk at checking things out. I have build in a security which warns if 100 synonyms are created for on phrase and which breaks if 500 are created.
If you check the result in the commandlist you can see that one synonym group triggers the other and you should select topmost only for one or both of them.In fact ‘topmost only’ is quite sufficient in most cases. It can be regarded as a placeholder that creates many phrases (one contains the placeholder).
Maybe ‘topmost only’ should be turned on by default.All phrases available for speech recognition, created as a result of synonyms and variable selections are shown when checking the display checkboxes.
Variables and synonyms are just features. If you want to you can create 20 thousand commands without using synonyms and variables.
-
V1.141012 released
- enable response issues fixed - thanks again to Wingedsky
- hotkey device manager rewritten
-
That seems to be near perfection now Livrot
Tip : if you want to use PPT, it is necessary to modify ambiguous and unambiguous timeout initial settings to be able to release PPT button before LMC stops speech recognition.
otherwise it generates a popout about lack of microphone. my settings are about 2s:rolleyes:Thanks a lot Livrot, this time after more than 3 hours of tests, modifying profile, …. no bugs and errors!
Beta version can go away! YEHHHI just need to be more confident with your" topmost" option : not sure to have understand perfectly how it works. But I will learn;)
I will post my profile as soon as it is completed and tested
Windblow
-
Thanks for the kind words, Wingedsky!
The explanations are often very brief and examples are missing from many tooltips. I wish I had the relaxed living situation to concentrate on pushing LMC forward to the market.
But for now I really am missing the air to breath here in Germany…Videos are planned but maybe next year if ever
Maybe it helps you if I just tell what the synonym recursive function does:
The recursive function is called when display is switched on or when enable is selected and the user clicks OK when leaving the editor.
One phrase in the commandlist after the other is put in an empty list of strings (at the beginning of the list)
and the recursive function (RF) is called with this list as parameter.
The returned list contains all phrases resulting from the synonym creation and the original phrase still at the beginning.When the RF gets the list it checks all synonym groups if one synonym is found inside the phrase. (With topmost selected only the first element of the group is searched for.)
If a synonym is found all the other synonyms of that group will create new phrases replacing the found synonym.
These new phrases are added to the list of strings.
If the list of strings has increased in size the RF calls itself again - the new phrases will be checked for other synonyms, too.
If the list of strings did not increase, the list is returned as return parameter. -
Thanks for the explanation Livrot. As I can understand the principle, I have to measure the consequences when editing the profile.
I have uploaded my profile on your site , based on yours.At the moment, texts are there most of the time to control the correct recognition.
in Future version I will change some texts and add phrases with only texts in responses with no action. Just for better immersion I hope.
Don’t forget to set “use profile settings”
Enjoy:D
Windblow -
Thank you very much for your contributions to LMC, Wingedsky!
The website will permit .zip file uploading soon in order to distribute the soundfile package and maybe explanatory pdf file / screenshots, too.
-
Livrot, do you think it would be possible to set variables in texts part ?
for example :
phrase : __callsign flightnumber Uniform check
text1 : __callsign _flightnumber wingmenthis would be great
another suggestion : the possibility to choose the number of texts. By default it is 3, You Wrote that it could be easy to increase the number. Could you make a box to select the number we desire?:rolleyes:
Windblow
-
Both can be done, but I need a break from LMC development for three weeks to arrange RL issues.
Livrot
-
No Problem Livrot, LMC is perfectly operationnal now and is running quite good;;)
It is just an idea for Livrot Mic Command V1.141225 version !!!:rolleyes:
I hope things will be fine for you :(:)
Windblow
-
Just a heads up:
My example profile for variable / synonym use is incomplete and partially wrong (output) - It was just a conversion of an old ‘shoot!’ profile for maybe another Falcon Version.
Please convert your own VAC / shoot / keyspeech profiles or check out Wingedsky’s profile.Also:
You might have to set a delay on the plain text output (30ms) else artificial keypresses are send out so fast that they get lost in BMS input recognition (twice the same regarded as one etc.)
use the ‘apply globally’ checkbox under the delay slider to use the selected delay for all plain text output.have a nice flight
-
And As my profile was based on yours, I also have detected some errors in mine.:rolleyes:
So I will correct that as soon as possible.
But anyway, it can do the job at 90%, not so badWindblow
-
Hi Guys,
here are my profiles for Livrot Mic Command V1.141012:-
with text actions :Windblow-LMC_BMS_4.32_text_profile_V0.9beta
-
converted with makro actions : Windblow-LMC_BMS_4.32_makro_profile_V0.9beta
-the sound file (be conscious that you will have to adapt the path) : LMC_recogn
Don’t forget to set “use profile settings” before loading the profile.
in this version, except a few phrases, texts in responses are set to verify speech recongnition at this stage of development. You can disable all sounds if you feel it better…
Now the profile is good at 99% I think.
Livrot, feel free to erase my previous one on your site. For these version, maybe it is better to wait a few day before uploading it
Let me know if you find some errors in commands.I hope it will help
Enjoy
Windblow -
-
Haha, you found the errors last time. Didn’t thank you for that, sorry.
I feel I can recommend your profile without any test.
-
Hi Livrot,
here is another suggestion in addition with previous in post #148,Would it be possible to implement the possibility to play an action in response, as it is possible to put a sound file in text ?;)
Example :
phrase : Kunsan approach viper 11 entering in the hold
Action : none
text 1 : Viper 11 kunsan approach, continue holding
text 2 : Viper 11 kunsan approach, copy
text 3 : tt&
text 4 : tt’
response : #r#1#2#a3#a4#rthus, response would be randomized, with a “copy”, or with wind information or qnh or what ever you can imagine …
I hope your personal things are working good:)
Windblow
-
good idea, noted.
A workaround for this exists (complicated, but without random):
same as above but with action set to a makro like CTRL SHIFT BACKSPACE
and an additional command with CTRL SHIFT BACKSPACE as hotkey and output ‘tt&’ + delay as needed (if hotkey self triggering works already - haven’t checked yet)
same for >tt’<
It is noted and easy to do. Thanks for another good idea, Wingedsky!
I’m back at it and working on a 99 texts per command version so this is the time to add ideas for the response parser if anyone has another idea yet.
These 99 lines are editable like a text with linenumbers (1-99) and can be saved to .txt and loaded from .txt files.
Line 1 - 3 (= text1, text2 , text3) are the only ones shown in the commandlist.Right click into this textbox gives a menu allowing to browse for a soundfile or to load / save to .txt plus the usual cut, copy, paste, undo, redo, select all, select line, clear all.
Also if a line contains just a xxx.wav (xxx being the name of a soundfile) this file would be searched in the folder where the profile was loaded from, allowing to share the folder as a working solution with everything in place.
A text/response containing ‘.wav’ indicates a soundfile, this way the #s1 could be just #1 in the response string.
The ‘is soundfile’ checkbox is gone to unclutter the response box (not needed any more, too). -
:rolleyes:hum, Another request for me Livrot :rolleyes:: please try to make profiles edited with V1.141012 compatibles with the future releases.
It would be very cool !
-
no problem.
upcoming new response commands:
#a34 = action in text 34 -> text34 equals a phrase in the commandlist and the concerned command will be executed (plain text / run file or makro)
#d500 = wait 500 ms
#d500-2000 = wait minimum 500 ms and up to 2000 ms (per random)