KnowBrainer Speech Recognition
Decrease font size
Increase font size
Topic Title: Trigger word for instruction versus dictation
Topic Summary: Trigger word for instruction versus dictation
Created On: 06/14/2020 03:41 AM
Status: Post and Reply
Linear : Threading : Single : Branch
 Trigger word for instruction versus dictation   - MiltonK - 06/14/2020 03:41 AM  
 Trigger word for instruction versus dictation   - kkkwj - 06/14/2020 01:29 PM  
 Trigger word for instruction versus dictation   - R. Wilke - 06/14/2020 01:43 PM  
 Trigger word for instruction versus dictation   - MiltonK - 06/16/2020 11:44 AM  
 Trigger word for instruction versus dictation   - Lunis Orcutt - 06/16/2020 08:14 PM  
 Trigger word for instruction versus dictation   - PG LTU - 06/14/2020 02:35 PM  
 Trigger word for instruction versus dictation   - Lunis Orcutt - 06/14/2020 02:49 PM  
 Trigger word for instruction versus dictation   - Ag - 06/16/2020 03:23 PM  
 Trigger word for instruction versus dictation   - R. Wilke - 06/16/2020 03:37 PM  
 Trigger word for instruction versus dictation   - R. Wilke - 06/14/2020 02:51 PM  
 Trigger word for instruction versus dictation   - Ag - 06/15/2020 05:36 PM  
 Trigger word for instruction versus dictation   - MiltonK - 06/16/2020 11:46 AM  
Keyword
 06/14/2020 03:41 AM
User is offline View Users Profile Print this message

Author Icon
MiltonK
New Member

Posts: 15
Joined: 10/19/2014

Many years ago (before Dragon arrived) there was a voice to type program available called Via Voce.  Although Dragon is obviously way better than that product the cool thing about that product is that one could establish a trigger word which the programme would recognise as being a precursor to an instruction as opposed to normal dictation.  It is such a pity that Dragon doesn't have the same facility as I really struggle to get Dragon to recognise an instruction and distinguish it from dictation.  I was wondering if one of the boffins have some magical way of introducing this feature to Dragon.  It could be a word or a keystroke but either way would be a fantastic additional to.  I will keep a lookout for a response..



-------------------------
Milton Koumbatis
 06/14/2020 01:29 PM
User is offline View Users Profile Print this message

Author Icon
kkkwj
Top-Tier Member

Posts: 497
Joined: 11/05/2015

I don't think you can do that easily for Dragon builtin commands, but people sometimes define their own prefix keyword to their commands (for example, Edgar uses "zotz" and Ag uses "puff"). I imagine that you could define your own command like "zotz " or "puff " and then have your script take each word in the part and inject it into a HeardWord command. That might work.

For example, "zotz move to top of document" would end up as

Heardword "move","to","top","of","document"

inside the script.



-------------------------

Win10/x64, AMD Ryzen 7 3700X, 64GB RAM, Dragon 15.3, SP 6 PRO, SpeechStart, Office 365, KB 2017, Dragon Capture, Samson Meteor USB Desk Mic, Klim and JUKSTG earbuds with microphones



 06/14/2020 01:43 PM
User is offline View Users Profile Print this message

Author Icon
R. Wilke
Top-Tier Member

Posts: 7229
Joined: 03/04/2007

That particular feature has always been built-in. Press the control key to activate command recognition only. Alternatively, press the shift key to activate dictation recognition only.

No need to go down to the levels of creating custom commands and prefixing them with your own stuff, unless you are geared for fiddling.



-------------------------



No need to buy if all you want to do is try ...

DragonCapture KB Download (Latest)
DragonCapture Homepage

 06/16/2020 11:44 AM
User is offline View Users Profile Print this message

Author Icon
MiltonK
New Member

Posts: 15
Joined: 10/19/2014

Wow I didn't know that it was already built in!  So if I click the control button once everything I say thereafter will be recognised as a command.  If I click shift once then every thing I say will be understood to be a command?  I just tried it but it didn't seem to work.  Must I hold the key down or is just one click okay?



-------------------------
Milton Koumbatis
 06/16/2020 08:14 PM
User is offline View Users Profile Print this message

Author Icon
Lunis Orcutt
Top-Tier Member

Posts: 37997
Joined: 10/01/2006

Originally posted by: MiltonK Wow I didn't know that it was already built in!  So if I click the control button once everything I say thereafter will be recognised as a command.  If I click shift once then every thing I say will be understood to be a command?  I just tried it but it didn't seem to work.  Must I hold the key down or is just one click okay?

 



You can only lock Dragon temporarily into Command Mode or Dictation Mode when you press and hold the {Ctrl} Key (command) or {Shift} Key (dictation).



-------------------------

Forum Mission Statement
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ext 1



 06/14/2020 02:35 PM
User is offline View Users Profile Print this message

Author Icon
PG LTU
Top-Tier Member

Posts: 2072
Joined: 03/21/2007

+1 Right on Rüdiger. PS a few of us here started with Via Voice . . .


Or a command "Prefix (mine is "Doris") [dictation]" which changes the recognition mode to Commands, emulates recognition of [dictation] as a command, and turns the mode back to Normal, all quick as can be.

One can even imagine a profile having but a single command as outlined above except rather than emulating the [command name from the dictation variable] (which necessitates the command be in the User Profile) one instead shellexecutes out to the command set or '#Uses comments it from another set of Dragon commands without the prefix. 


Hth, but see below



-------------------------




PG





Remember folks, my comments and this forum are for entertainment value only, please, no wagering or other reliance on the contents herein.  I permit no commercial use of my ideas (whether expressions or embodiments) without my written consent.



 06/14/2020 02:49 PM
User is offline View Users Profile Print this message

Author Icon
Lunis Orcutt
Top-Tier Member

Posts: 37997
Joined: 10/01/2006

History:

 

Bell Labs invented speech recognition in 1952, not arguably counting Thomas Edison's 1877 recorder. IBM picked up the ball with dedicated hardware speech recognition in 1962, which recognized 16 words. In 1976 Harpy, by Carnegie Mellon, could recognize 1011 words. Texas Instruments, the people who invented the LED calculator, introduced Speak & Spell which was based on a dedicated chip utilizing hidden Markov modeling. In 1992 Apple created the 1st continuous speech recognition capability with a 20,000 word vocabulary but DragonSystems was the 1st company to introduce a software release of non-continuous speech recognition (one word at a time) in 1990, for $9000. DragonSystems continuous speech was released in 1997 and we were likely the 1st third-party NaturallySpeaking beta testers in 1996. DragonSystems even gave us a state-of-the-art notebook computer to Mobley work with. If anyone is interested, the 1st KnowBrainer command utility was created in 1993 and went public when DragonSystems marketed NaturallySpeaking (continuous speech recognition) in 1997. A year later, thinking that speech recognition would take off, we secured the www.KnowBrainer.com website.


ViaVoice was introduced a few years after Dragon NaturallySpeaking which competed with Kurzweil (briefly), L & H (Lernout & Hauspie) and Phillips. Nuance discontinued using the original speech engine which was introduced in 1997; in favor of the Philips (based in the Netherlands) engine which explains why Dragon 15 was originally released in English and German; a 1st.


ScanSoft purchased ViaVoice in a stock swap. Nevertheless, ViaVoice was always trailing Dragon (too little too late) but we personally liked their ability to create step-by-step commands and post alter the deployment speed as much as 10 times slower.

Various Dragon owners debated the concept of adding 1 or 2 syllable trigger commands and even the possibility of using separate commands to switch from Dictation Mode to Command Mode (which was briefly utilized in 1997) but decided that neither option was best. In 1998 DragonSystems decided to make Dragon dictation and command deployment simultaneous. In some cases, like new paragraph, both modes can be combined without pausing. Although we don't approve of a trigger command, we deliberately made the KnowBrainer XML command file end-user editable so that if you want to simultaneously add any trigger word, to all KnowBrainer commands, it could be accomplished in under 10 seconds.

Of course as Rüdiger noted, arguably the easiest approach would be to hold the {Ctrl} key down to temporarily lock Dragon into Command Mode and the {Shift} key to lock Dictation Mode.



-------------------------

Forum Mission Statement
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ext 1



 06/16/2020 03:23 PM
User is offline View Users Profile Print this message

Author Icon
Ag
Top-Tier Member

Posts: 449
Joined: 07/08/2019

I enjoyed Lunis's history of speech recognition, but some  of it was inconsistent with what I remember, and also with  the IBM webpage below and Wikipedia

 

https://www.ibm.com/ibm/history/ibm100/us/en/icons/speechreco/breakthroughs

 

1962 - IBM shoebox 16 words Seattle World's fair

 

1984  first real-time large vocabulary dictation system demonstrated by IBM speech  time,  needed IBM minicomputer and three array processors, a full room of hardware

 

1987 ported to special-purpose hardware on IBM PC AT

 

1992  IBM released first dictation system, the IBM speech server series

 

1993  IBM personal dictation system, first dictation system for  personal computers.  later renamed IBM VoiceType  dictation. 32,000 words at a rate of approximately 70 to 100 words per minute, with 97 percent accuracy.

 

1996 IBM VoiceType SimplySpeaking,  with Microsoft Windows applications 22-42K  word vocabulary.  continuous speech. E.g. radiology

 

1997  IBM introduced ViaVoice -  continuous dictation in multiple languages

 

Continuing in Wikipedia https://en.wikipedia.org/wiki/IBM_ViaVoice

 

1999, IBM released a free of charge version of ViaVoice.

 

In 2003, IBM awarded ScanSoft, which owned the competitive product Dragon NaturallySpeaking, exclusive global distribution rights to ViaVoice Desktop products for Windows and Mac OS X. Two years later, Nuance merged with ScanSoft.[8]

 

(This is what threw me in Lunis's history -  I may have misread it, but he seems to suggest that ViaVoice came from ScanSoft, not IBM.)

{Dragon misrecognitions continue to surprise:  "Scansoft" as "San Pablo"??  Although the IBM history of speech recognition provides insight: instead of trying to recognize sounds as the human ear and brain might, they match based on more arbitrary features.}

 

== 

 

 I used IBM VoiceType, SimplySpeaking, ViaVoice, as well as early versions of Dragon

 

 I think I remember that  Dragon's  first commercial  product was non-continuous commands, while at the time IBM could do the continuous dictation.   I remember trying to switch back and forth between the two.



-------------------------

DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design.



 06/16/2020 03:37 PM
User is offline View Users Profile Print this message

Author Icon
R. Wilke
Top-Tier Member

Posts: 7229
Joined: 03/04/2007

I enjoyed Lunis's history of speech recognition, but some  of it was inconsistent with what I remember, and also with  the IBM webpage below and Wikipedia

 

The department of history and all things Wikipedia at the Trump university was severely underfunded.

 

 



-------------------------



No need to buy if all you want to do is try ...

DragonCapture KB Download (Latest)
DragonCapture Homepage

 06/14/2020 02:51 PM
User is offline View Users Profile Print this message

Author Icon
R. Wilke
Top-Tier Member

Posts: 7229
Joined: 03/04/2007

Great idea, PG, except for "Doris" being what one of my aunts was called, so I'd rather not go anywhere near it.

My overall suggestion: If you can avoid fiddling by any means, better do so.




-------------------------



No need to buy if all you want to do is try ...

DragonCapture KB Download (Latest)
DragonCapture Homepage

 06/15/2020 05:36 PM
User is offline View Users Profile Print this message

Author Icon
Ag
Top-Tier Member

Posts: 449
Joined: 07/08/2019

I am probably one of the biggest advocates of having a "prefix" for most commands.

I have edited MyKBCommands.xml to add "puff" to nearly all standard KnowBrainer commands, and "PUFF" to nearly all of the commands that I added myself.

("Nearly all" and not "all" because, per Lunis's recommendation, I have utterances like "Hello" and "Hello Siri" un-prefixed, somewhat reducing the problems that happen when I talk to my wife, or answer the phone. I also have "Close Page" and "Close Window" unprefixed, although in this case they just print a MsgBox error saying "say puff close page" instead. Close page/window IIRC what inspired me to use the prefixes. "Space Bar". "Resetart Dragon" - another error message box, saying "if you can see this then SpeechStart+ is not running" (although usually I get both this and the SpeechStart+ message). BTW Since I started wearing a wristband with a button tied to microphone off/on, I find that I don't need the "Hello" un-prefixed voice commands as much)

I get great pleasure out of saying "puff", as in "Puff the Magic Dragon".

Although I do tend to have problems like "puff" being confused with "off" and a few other words. Also, plosives like "p" being to frequently be noise filter, as one might expect.

Edgar's recommendation of "zotz" as the prefix seems to have fewer misrecognitions than "puff". but "puff" makes me smile! I wish I knew of a dragon name from some famous story that was one syllable but not frequently confused they word or syllable that I use in ordinary language.

--

Unfortunately, this prefix only applies to KnowBrainer commands, which can be edited. It does not apply to standard Dragon commands, KnowBrainer's hardcoded commands like "EDIT Selected Command", SpeechStart+ commands like "RESTART Dragon", or "DISPLAY Middle", or DragonCapture commands like "show box". you can define your own prefixed KnowBrainer commands that invoke some of those commands, you cannot eliminate their un-prefixed form. which, for me, was a primary reason that I added these prefixes.

Nevertheless, adding these prefixes dramatically reduces the number of accidental invocations of KnowBrainer commands.

---

Hand editing MyKBCommands.xml is possible, and might be a good way to start, but it makes it harder to import Lunis's latest version of the KnowBrainer commands.

I use Perl scripts to manipulate MyKBCommands.xml. I will be happy to share such scripts, although I blush to admit that at the moment they are not bulletproof. actually, when I started I was sloppy and editing, and I need to clear that up. I haven't upgraded KnowBrainer's commands in a few months.


---

BTW you can also add suffixes to any command that does not have open-ended argument lists. Unfortunately, some of the most useful commands have parameters, so I can't recommend suffixes, even though I often find them easier to say. overall, I am noticing a pattern whereby I add vocabulary as suffixes or infixes but commands with prefixes - e.g. I prefer to say "5 as number" than "numeral 5". but treating commands and vocabulary differently violates good principles of UI design Or software modularity - it should be possible to move the implementation of an utterance from vocabulary to command arbitrarily.

===

As for RĂ¼diger at al's recommendation to press control for command mode and a shift for dictation mode (and even better, some other key for " press to talk") - That's great if you're near a keyboard. I sometimes use it when I am Sitting at my keyboard. However, frequently I am far away from my keyboard, so that particular technique for me. Or at least, it won't fly for me, until I get a Bluetooth button that I can wear on my wrist that provides shift/ctrl/whatever key up/down events.

===

BTW: Meta-observation

many people recommend command syntax of the form "verb-object". E.g. "open app", "close window". "Go to top of file".

adding a prefix makes it SVO, subject-verb-object, which happens to be English is standard form. "Puff close window".

but because the prefix makes things longer, and longer makes things less reliable for recognition, instead of saying "PUFF goto title" I changed my commands to "PUFF title", and so on.

I also tend to say "PUFF text red", or "PUFF table insert above/below/left/right" rather than "PUFF insert row above/...". although that is mainly so that similar commands are together in the KnowBrainer sidebar. I have written scripts to generate multiple Equivalent commands, like
"PUFF insert row above/below" ==> "PUFF table insert above/below", but I am not using them since I have had so many problems with stalls that I suspect that I might simply be overloading KnowBrainer/Dragon wrt number of commands, loading and unloading grammars. I'm not 100% sure, but I think that many of the stalls that I get in Dragon are related to changing applications/windows, and hence loading and unloading grammar sets.

I would be happy to say "PUFF insert above/below/left/right" if there were a reliable way of telling whether the cursor and/or the amount is in a table or not. But as far as I know there is no such reliable way, and as I'm sure most of us know blindly inserting a command for one context into a different context leads to bad things happening. I find that "PUFF table insert above/..." leads to fewer such accidents. It also allows me to have commands like "PUFF table help" to display help, a list of what commands are available. (TBD: automate generating such help, the way emacs does.) I have occasionally done "PUFF table" ==> help, which is even better when I can't remember what commands are available, although unfortunately that often leads to "PUFF table insert above" being heard as two separate utterances "PUFF help" and "insert above".



-------------------------

DPG15.6 (also DPI 15.3) + KB, Sennheiser MB Pro 1 UC ML, BTD 800 dongle, Windows 10 Pro, MS Surface Book 3, Intel Core i7-1065G7 CPU @ 1.3/1.5GHz (4 cores, 8 logical, GPU=NVIDIA Quadro RTX 3000 with Max-Q Design.

 06/16/2020 11:46 AM
User is offline View Users Profile Print this message

Author Icon
MiltonK
New Member

Posts: 15
Joined: 10/19/2014

Thank you so much to everyone for their replies and contributions.  Some of the stuff is beyond my computer skills but I will see what I can get right.  The history post was particularly interesting.



-------------------------
Milton Koumbatis
Statistics
31793 users are registered to the KnowBrainer Speech Recognition forum.
There are currently 0 users logged in.
The most users ever online was 12124 on 09/09/2020 at 04:59 AM.
There are currently 744 guests browsing this forum, which makes a total of 744 users using this forum.

FuseTalk Standard Edition v4.0 - © 1999-2020 FuseTalk™ Inc. All rights reserved.