[RESOLVED] VR V4.3 - Using Voice Controls
Posted: Sun Jul 01, 2018 2:25 pm
Congrats LM on including voice controls (experimental) in v4.3
(see Using Voice Controls via Prepar3D Learning Center)
I've been playing with voice control of P3D for many years and have a large vocabulary of almost all controls for the sim via VoiceAttack.
I'm just starting to play with what you have implemented currently and I like it a lot!
Early feedback/thoughts:
The three part csv VoiceControlPhrases.cfg is cool ([phrase], [event], [certainty]) but seems to be limited to one activation/switch of any particular Event ID so if the user needed to alter any incremental ID e.g. KEY_COM_RADIO_FRACT_DEC (Increments COM by 25 KHz) by more than one activation/switch then how may that be implemented? Only by use of the PDK? How can "ordinary Joe" get a better understanding of this to get to use it?
There are few Event IDs that are simply toggles with probably the most likely voice activated one being KEY_GEAR_TOGGLE. However, we can't associate a voice command to the 2 states of "lower gear" or "raise gear" when we don't know that the command will give us the desired result. Can KEY_GEAR_SET be used (Sets gear handle position up/down (0,1)). If so can you please give an example of how we might achieve this? PDK again?
I don't use ATC that often but I think there is some sort of activation issue of ATC anyway in v4.3 and FYI the ATC control line of:
open ATC, ATC, .93 probably should have been
open ATC, KEY_ATC, .93
but.... this is a toggle EVENT ID so closing it would be?
I seem to have recognition issues with this as the reliability of this command is awful?? Perhaps "open A T C" with spaces, dunno?
It would be good to have some optional audio feedback from voice commands too on a command by command basis e.g. a small beep or a user defined response such as "yes captain", "confirmed" or a simple "do it yourself ya lazy *um". Note to repeat the command back isn't a good idea on an open sound system. Once had my gear bouncing up n' down for hours -:).
I added the following as custom commands to just see it working:
key panel 1,KEY_PANEL_1, .9
key panel 2,KEY_PANEL_2, .9
key panel 3,KEY_PANEL_3, .9
key panel 4,KEY_PANEL_4, .9
key panel 5,KEY_PANEL_5, .9
key panel 6,KEY_PANEL_6, .9
key panel 7,KEY_PANEL_7, .9
key panel 8,KEY_PANEL_8, .9
key panel 9,KEY_PANEL_9, .9
all lights toggle, KEY_ALL_LIGHTS_TOGGLE, .9
gear toggle, KEY_GEAR_TOGGLE, .9
Sorry to report that NONE of the above works??
I'd like to ask if your speech recognition is reliant on the inbuilt Windows speech engine as is VoiceAttack?
If so, I can advise that this is very very poor in comparison to that which Nuance use in Dragon Naturally Speaking.
It seems VoiceAttack are happy to not use a better engine due to the costs that would have to be passed onto the end user.
(see Using Voice Controls via Prepar3D Learning Center)
I've been playing with voice control of P3D for many years and have a large vocabulary of almost all controls for the sim via VoiceAttack.
I'm just starting to play with what you have implemented currently and I like it a lot!
Early feedback/thoughts:
The three part csv VoiceControlPhrases.cfg is cool ([phrase], [event], [certainty]) but seems to be limited to one activation/switch of any particular Event ID so if the user needed to alter any incremental ID e.g. KEY_COM_RADIO_FRACT_DEC (Increments COM by 25 KHz) by more than one activation/switch then how may that be implemented? Only by use of the PDK? How can "ordinary Joe" get a better understanding of this to get to use it?
There are few Event IDs that are simply toggles with probably the most likely voice activated one being KEY_GEAR_TOGGLE. However, we can't associate a voice command to the 2 states of "lower gear" or "raise gear" when we don't know that the command will give us the desired result. Can KEY_GEAR_SET be used (Sets gear handle position up/down (0,1)). If so can you please give an example of how we might achieve this? PDK again?
I don't use ATC that often but I think there is some sort of activation issue of ATC anyway in v4.3 and FYI the ATC control line of:
open ATC, ATC, .93 probably should have been
open ATC, KEY_ATC, .93
but.... this is a toggle EVENT ID so closing it would be?
I seem to have recognition issues with this as the reliability of this command is awful?? Perhaps "open A T C" with spaces, dunno?
It would be good to have some optional audio feedback from voice commands too on a command by command basis e.g. a small beep or a user defined response such as "yes captain", "confirmed" or a simple "do it yourself ya lazy *um". Note to repeat the command back isn't a good idea on an open sound system. Once had my gear bouncing up n' down for hours -:).
I added the following as custom commands to just see it working:
key panel 1,KEY_PANEL_1, .9
key panel 2,KEY_PANEL_2, .9
key panel 3,KEY_PANEL_3, .9
key panel 4,KEY_PANEL_4, .9
key panel 5,KEY_PANEL_5, .9
key panel 6,KEY_PANEL_6, .9
key panel 7,KEY_PANEL_7, .9
key panel 8,KEY_PANEL_8, .9
key panel 9,KEY_PANEL_9, .9
all lights toggle, KEY_ALL_LIGHTS_TOGGLE, .9
gear toggle, KEY_GEAR_TOGGLE, .9
Sorry to report that NONE of the above works??
I'd like to ask if your speech recognition is reliant on the inbuilt Windows speech engine as is VoiceAttack?
If so, I can advise that this is very very poor in comparison to that which Nuance use in Dragon Naturally Speaking.
It seems VoiceAttack are happy to not use a better engine due to the costs that would have to be passed onto the end user.