Custom Commands not working when transcribing a file
Submitted by Art222 on Mon, 08/28/2006 - 19:33.
I'm writing a custom application using DNS SDK. The simple example below is for recognizing street addresses on the left and right side of street. A sample monolouge might be something like "Left 1, Right 2, Left 3, Right 4".
In my code, I have a list of directions (Left, Right) and a list of numbers from 1 to 9. Code snippet is below.
list.Clear();
list.Add("1");
...
list.Add("9");
VMenu.ListSetStrings("digits1_9", list);
list.Clear();
list.Add("left");
list.Add("right");
VMenu.ListSetStrings("direction", list);
VMenu.Add(2, " ", null, null, null, null);
This works great during normal interactive mode, but I am having trouble if I attempt the same thing using a dictation object by adding the following line.
axDgnDictEdit1.TranscribeFile ("test04.wav");
When I try to use my commands from above when transcribing, the transcribing doesn’t recognize my commands. Is DNS supposed to be able to recognize commands while in dictation mode?
Art222 wrote: I'm writing a
I'm writing a custom application using DNS SDK. The simple example below is for recognizing street addresses on the left and right side of street. A sample monolouge might be something like "Left 1, Right 2, Left 3, Right 4".
In my code, I have a list of directions (Left, Right) and a list of numbers from 1 to 9. Code snippet is below.
list.Clear();
list.Add("1");
...
list.Add("9");
VMenu.ListSetStrings("digits1_9", list);
list.Clear();
list.Add("left");
list.Add("right");
VMenu.ListSetStrings("direction", list);
VMenu.Add(2, " ", null, null, null, null);
This works great during normal interactive mode, but I am having trouble if I attempt the same thing using a dictation object by adding the following line.
axDgnDictEdit1.TranscribeFile ("test04.wav");
When I try to use my commands from above when transcribing, the transcribing doesn’t recognize my commands. Is DNS supposed to be able to recognize commands while in dictation mode?
First, it isn't clear from your explanation whether or you are using the full SDK (i.e., SDK purchased from Nuance independent of NaturallySpeaking) or you are using the SDK as implemented from within DNS Professional. You need to clarify this.
However, regardless of how you're developing your custom application, when transcribing from a wav file there is a "Commands" option, which you can find either under the "Transcribed from" dialog "Advanced Settings" (Sound | Transcribe Recording) or from the transcription commands in the SDK itself, which sets transcription to "All Commands." Unless this is set, Dragon NaturallySpeaking automatically defaults to "Restricted Commands." Under the latter condition your script will not be recognized.
I'm not sure that it will be recognized anyway given what you're trying to do because there are a number of variables that can preclude your script executing properly and/or completely. Recognizing all commands issued from with in a digital recording or wav file has always been somewhat problematic. Most of the problems stem from how clearly commands issued from with in such are issued (i.e., pauses before and after), as well as timing issues that are critical to the processor speed and rate of dictation playback. However, unless you have the commands recognition set to "All Commands" it won't work in any case.
This is all I can help you with without further clarification on exactly what it is that you are doing and howl you are doing it.
Chuck Runquist
Former DNS SDK & Senior Technical Solutions PM for DNS
Thanks Chuck, Your answer
Thanks Chuck,
Your answer was dead-on. I'm using DNS 8.1 SDK as implemented from within DNS Professional. Since the SDK and Professional can't be on same machine at same time, I've been flip-flopping. I was able to use the dragon bar to test your advice and it worked great. Now I just need to figure out what the command is programmatically since my help files got knocked out when I flipped from SDK to Professional.
Art.
I can't find any information
I can't find any information on this subject (or many other subjects) in the SDK nor API help documentation. Intuitively, most of the community is focused on the user side and it seems as though Scansoft is rather light on documentation for developers. Is there another source or book that I can purchase to learn more from the developers perspective?
Chuck, I always value your
Chuck,
I always value your advice, but I don't think you got 100% DNS accuracy on your answer:
"howl you are doing it."
I think it's a great voiceo, but why DNS would think you are saying "howl" is beyond me!
Matt
Matt Chambers
Chuck,
I always value your advice, but I don't think you got 100% DNS accuracy on your answer:
"howl you are doing it."
I think it's a great voiceo, but why DNS would think you are saying "howl" is beyond me!
Matt
Matt,
At least everyone knows that I am dictating.
Two reasons: (1) I was tired and obviously not enunciating clearly, and (2) I live in Howell, New Jersey, so when I slur or run my words together it probably picks up the homophonic commonalities. I think what I was saying at the time was "how are you doing it." It probably makes more sense that the misrecognition was based on slurring that phrase/sentence.
Chuck Runquist
Former DNS SDK & Senior Technical Solutions PM for DNS
Chuck Runquist wrote: ...so
...so when I slur or run my words ...
I don't think that Matt intended it to be a slur on your reputation
Quentin
crivon1 wrote: Chuck
...so when I slur or run my words ...
I don't think that Matt intended it to be a slur on your reputation
Quentin
Quentin,
Sorry if my post was misinterpreted by anyone. It was intended to be funny.
No offense was taken.
Part of that was posted because I have stated for more than 10 years that it is mathematically and practically impossible for SR to ever be 100% accurate. Human speech between 2 people is never 100% accurate, how can computers be such?
Chuck Runquist
Former DNS SDK & Senior Technical Solutions PM for DNS
Hi Chuck, Not to seem stick
Hi Chuck,
Not to seem stick in the mud, but I think my question may have been lost in the conversation detour.
I can't find any information on this subject (or many other subjects) in the SDK nor API help documentation. Intuitively, most of the community is focused on the user side and it seems as though Scansoft is rather light on documentation for developers. Is there another source or book that I can purchase to learn more from the developers perspective?
Thanks again for all your help.
Art.
Chuck, My comment was
Chuck,
My comment was intended to be punny
Quentin