Windows™ Speech Recognition Toolkit
Contents
1. the text surf the net appears in the When Say textbox You say find program A file browsing window opens up that will allow you to point to the program you d like to run You say mozilla firefox The directory will change to Mozilla Firefox Note this may be different on your computer You say firefox firefox exe is selected and the full path C Program Files Mozilla Firefox firefox exe is placed in the Add A Program to Run textbox You say add program The Add a Program to Run textbox is cleared and the Command Sequence Viewer displays the text RUN C Program Files Mozilla Firefox firefox exe More on the Command Sequence Viewer shortly You say create macro with command sequence That s it you ve created a Command Macro that will open the Firefox browser anytime you say surf the net Easy as 1 2 3 right Well almost Here s a screenshot with the four different areas highlighted Text Macros Command Macros About Command Macros Allows you to create multiple step by step commands using common keyboard shortcuts In The Context Of Global eposions When Say surf the net Simulate Keypress F ALT Command Key E CTRL SHIFT Numbers Letters Words Keystroke Preview Next we ll create a practical Command Macro that will only work in the context of Firefox and will do the foll2. Command Macros This tab is possibly the most exciting of all within this application It allows you to control your computer with voice commands that up until now have only been available with the keyboard and mouse Don t be intimidated by all those buttons and textbox areas on this tab We ll cover them thoroughly one by one When we re done you ll confidentially be creating sophisticated macros that do some pretty amazing stuff Here s the screenshot for this tab Text Macros Command Macros About Command Macros Allows you to create multiple step by step commands using common keyboard shortcuts In The Context Of Gob Mogan gt Command Sequence Viewer When Say Simulate Keypress FJ CTRL We ve already covered the In The Context Of and the When Say textbox areas so if you re unsure about their use just click the particular hyperlinked words to go there and read about them Let s create a real world example two actually that demonstrate the many features on the Command Macros tab The first macro we want to create will listen for the words surf the net and responds by opening the Firefox browser We ll keep the context global because you ll want this macro to work no matter what you re doing or what program you re currently working in You say When say The cursor moves to the When Say text box area You say surf the net
3. This will produce a prompt box like this Please enter your name The resultState ans will store the input from the user in the ans variable that can be used in your macro Note you can name the resultState variable anything you like Two similar tags the alert and confirm nn lt alert title timeout gt lt confirm title timeout gt lt waitFor gt Allows you to pause macro execution lt waitFor seconds 25 gt Other acceptable attributes include anyStateNamelsSet and allStateNamesAreSet but the seconds attribute is most often used lt emulateRecognition gt Exactly as the tags name implies it allows you to emulate a spoken command lt emulateRecognition gt stop listening lt emulateRecognition gt Would send the recognizer the text stop listening just as if it had been spoken If you re still hungry for more don t worry There is plenty that hasn t been described in this brief overview of macro tags We ve only touched on the most common ones Check out some of the online groups and blogs There s a growing base of information available Train From Text The purpose of Train From Text is to improve the acoustic model of your user profile The Rainbow Passage is suggested as it was developed by linguists and contains all of the phonemes individual sounds the b in bat for example found in the English language If you wish to tra
4. Adds words to your personal speech dictionary with or without pronunciation for increased recognition accuracy 3 Add From File improves the language model of your speech profile by taking the text you pass it currently Word documents and text files parses it into words and then records the relative frequencies of occurrence of each word with respect to the words around it Professionals on the run requested a Transcription tool to transcribe wav files from a high quality digital recorder into text With our Transcription tool point to a wav file some recorders must convert from dss files to wav for transcribing into text easily copied pasted into another document Table of Contents ICPOGU CLON tostada REO Or tanta srendsaiede OE A E ER 2 Ta ble 0fCONtENES umd da A tala 3 Installing the WSR Toolkit c cccccccccssssssssececeeecsesessaeseceeeescsesseaeeeeeessesseseeaeeeeeeseeseeseaaeaeeeeeesseesees 4 TOXICO reia a toc Aa O A E A T 5 Command Macros arica dara dada dci ariel laca dando ita 8 Macro Editor s 16 Train tromi Text lindes 22 AA TO DINTO aT Vid E AE OE E EAE 23 RAN SCA PON a AA EA NAAA AA A ARA e AAA 25 Add Fron Fl ion cos 26 TIPS 8E TICKS A ahs saantcereih ce Nad Seeds lta ied aceon ats otek odie he ca EATE EE 28 Installing the WSR Toolkit After downloading the install file and executing it you ll be prompted for an install directory Once selected the install process will begin The Toolkit
5. a particular macro You may decide to post a question online and you can easily copy your macro code from here and paste it to a website for help The Open button will allow you to browse to your Speech Macros directory and select any file you macro file you wish Note Macro files end with the extension WSRMac The Save As button will allow you to save changes to your macro Careful though You should have a good idea what your doing before doing any macro editing The Cancel button clears any text in the macro textbox Exit will close the program For those wanting to know more here are some of the currently available XML tags available for Speech Macros along with a brief description Note these tags are case sensitive you must use exact upper lowercase as shown here lt speechMacros gt All macro files must use this tag and end with a lt speechMacros gt tag Only one set of these is allowed for each macro file lt command gt Required for every command and must be terminated with a lt command gt tag lt listenFor gt This tag is what is used by the macro facility to listen for what is being spoken It must be enclosed in a lt listenFor gt tag For instance lt listenFor gt surf the net lt listenFor gt When the computer hears the word or phrase contained within this tag it will execute everything within the preceding lt command gt tags lt sendKeys gt Using this tag yo
6. delay to allow the application to catch up From experience 25 seconds is a good delay for this type of thing You say delay increment The cursor moves to the Delay Increment numerical selection area Okay now for some fancy footwork We need to by voice control click the up arrow five yy tt times That will give us a delay of 25 seconds So you can either say press up press up y lt nt oress up press up press up OR simply say press up five times Pretty neat huh You say press up five times The Delay Increment shows 25 You say insert delay The Delay Increment area resets back to 0 00 and the Command Sequence Viewer displays a new line of text For the final sequence of our command macro we need to emulate the keypress ctrl c That copies whatever is highlighted to the clipboard so you ll be able to paste it somewhere else You say contol The CTRL checkbox is checked You say numbers letters words The cursor moves to the Numbers Letters Words textbox You say c The letter c is placed in the Numbers Letters Words textbox You say add The Keystroke Preview area clears and the Command Sequence Viewer displays a new line of text You should have three command sequences listed Now for a word about the Command Sequence Viewer text area This is the area in which as you buil
7. s necessary to accomplish your task you can simply insert them here in sequence with delays to accomplish your task via voice Don t know the keyboard shortcut Search our additional documentation for complete lists of the most popular applications keyboard shortcuts Macro Editor If you re curious about Windows Speech Recognition Macros and want to learn more you re in the right place On this tab you ll get to see the inner workings and details of the macro language itself If you re not interested in this stuff just skip over this section for now You can always come if you need to dig a little deeper Let s jump right in with a screenshot of the Firefox macro created in the Command Macros tutorial Windows Speech Recognition CAM Maco Etor About Macro Editor For advanced users and those who wantto learn more This tab allows you to manually create edit and review existing macros lt Command macro created with WSR d MacroBuilder 7 15 2008 8 01 lt speechMacros gt lt command gt lt condition operator not gt lt applsInForeground processName MacroBuilder exe gt lt condition gt lt listenFor gt surf the net lt listenFor gt lt run command C Program Files Mozilla Firefox Firefox exe params lt command gt lt speechMacros gt Don t worry about all that code for now Just know this is a great place to come if for some reason you re having a problem with
8. the previous paragraph But what if you d also like to create a macro in Firefox when you say back you ll emulate clicking the back button You can see the need for context Back could possibly mean different things in the context of different programs So what is the Global context The name pretty much explains it With Global All Applications selected your macro will work in all applications In other words the In The Context Of limits the functionality of your macro to a specific application So now the question is do you need to limit your macro to a particular application If so and the application you d like to confine your macro to is listed in the list just say the applications name Winword for Microsoft Word for instance If it s not listed in the dropdown list just say or click Browse Did you notice the underlined B in Browse As previously mentioned that means you can also press and hold the ALT key while pressing the letter b And just for fun you could also say press alt b You ll want browse to the location of the application you d like to choose click on the program name and it will automatically be put in the context textbox If after proofreading your Text macro you re ready to create it just say Save For reference the Cancel button allows you to clear the Text Macro tab and start fresh The Exit button will close the program
9. voice recorders save files as wav then choose the Browse button to point the transcription module to the wave file and choose Begin Transcription You ll soon see your words appearing in the transcription textbox Once the transcription is complete the entire text will automatically be selected and copied onto the clipboard ready for you to paste into a more suitable application Recognition Toolkit Text Macros Command Macros Macro Editor Train From Text Add To Dictionary i Transcribe From Wave File Using this feature will allow you to dictate to a portable voice recorder then after putting the wave file on your computer you can transcribe the file into text which can be copied and pasted into any application Note You ll get better transcription accuracy if you use Train From Text and Add From File before transcribing your file Choose Browse to locate the wave WAV file then choose Begin Transcription to begin i Browse Add From File This feature is best described as follows Add from File allows you to improve the accuracy of your Language Model It s not the same as adding words to your personal dictionary but works by taking the text that you pass it parses it into words and then records the relative frequencies of occurrence of each word with respect to the words around it If you use documents that represent your style of speaking this f
10. Windows Speech Recognition Toolkit A Speech Recognition Utility 2008 eMicrophones Inc Introduction Thank you for your interest in the Windows Speech Recognition Toolkit We hope this product will serve you well and that you will have many rewarding hours of hands free computing as a result of using its features The Windows Speech Recognition Toolkit was developed by eMicrophones Inc which has 15 years experience in the Speech Recognition industry Since previewing Windows Speech Recognition at Redmond in the summer of 2006 we thought very highly of the accuracy of the speech engine the ability to command and control one s computer and the forethought given to the graphical user interface However we realized some important features typical in other Speech Recognition software was missing The WSR Toolkit is meant to fill in the gaps and make a typical home or business user more efficient The Text Macro Builder easily allows you to create re usable boilerplate text The text is inserted when you issue the name you called the text block when creating it Three other features improve the accuracy of your personal user speech profile These are 1 Train From Text improves the acoustic model of your speech profile It allows you to train from text of your choosing We recommend you begin with The Rainbow Passage default text as it has an even distribution of the phonemes of the English language 2 Add To Dictionary
11. cessName notepad exe gt Allows you to restrict the speech macro command to a specific application You can also use windowlTitle and windowTitleContains attributes instead of the processName attribute So if you were interested in having a macro only execute when you had a specific document open in Word the tag might look something like this lt applsinForeground windowTitleContains MyDoc doc gt lt speak gt This tells the speech recognition system to talk back to you Whatever text you put between the speak tags will be audible on the speakers lt speak gt the voice box is a wonderful instrument lt speak gt lt setFeedbackText gt The following code lt setTextFeedback gt Hello World lt setTextFeedback gt Would produce this in the feedback text area lt switchToApp gt This tag allows you to switch focus to an already running application lt switchToApp windowTitleContains MyDoc doc gt would switch to your MyDoc document You can also use the windowTitle attribute here Along this same type of functionality there are the following additional tags lt minimizeApp gt lt maximizeApp gt lt restoreApp gt lt closeApp gt These also support the windowTitleContains and windowTitle attributes lt prompt gt Getting input from the user can be accomplished with the prompt tag like so lt prompt title Hello resultState ans gt Please enter your name lt prompt gt
12. d your Command Macro allows you to essentially oreview the sequence of commands before you actually create the macro It s a great tool for providing a visual double check of your work in progress While on the subject of the Command Sequence Viewer notice the Up Down and Remove buttons The Up and Down buttons will allow you select various lines in your Command Sequence The Up simply moves up the list the down highlights the next item in the list Once an item is selected you can say Remove and that particular sequence item will be removed from the list It s simply a way to edit your macro before saving it Now that you ve reviewed your command sequence in the Command Sequence Viewer You say create macro with command sequence And that s it you re done The macro has been saved to the Speech Macros directory and automatically loaded into the macro system which means it s listening for your command Go ahead try it Just remember since we set the context to firefox exe you ll have to have Firefox up as the active application With a little bit of creativity and thought you ll soon be creating voice command macros to do just about anything imaginable on your computer Got a question Still scratching your head about some of this Just post your question to one of the WSR macro forums and more than likely you ll have an answer in short order Here are a couple of good forums http ww
13. derstands you say it again if the recognition is still off just say spell it and you can spell it letter by letter You say Insert the text The cursor moves to the Insert The Text text box area You say John Smith press enter 1212N period north street press enter Greenfield comma o h press enter open parenthesis 5 5 5 close parenthesis 5 5 5 hyphen 1 2 1 2 press enter j smith at sign greenfield dot o h dot com press enter w w w dot j smith dot g o h dot com press enter The text will be inserted as you talk or type You may have to experiment a bit to get the proper recognition Now let s backup to the In The Context Of textbox You say In the context of The cursor moves to that dropdown box and several options are displayed Here s something similar to what you ll see Text Macros Command Macros Macro Editor Train From Text In The Context Of lobal Global All Applications Excel exe Firefox exe iexplore exe Winmail exe Notepad exe Outlook exe Winword exe set The Text In These are some of the more common applications you may be using Before deciding whether you ll need to select one of these Browse to find another or go with the default Global option some explanation on what is meant by context is in order Let s say you d like to create a macro for Microsoft Word in which you say back and it will automatically move you to
14. eature will improve dictation accuracy Choose Browse to find the document you wish to add then choose Begin Add from File The file will be opened and read into the language processor and quickly added to your language model Windows Speech Recognition Toolkit POS Transcription Add From Fie Add Text From File This feature will allow you to improve the accuracy of your Language Model It s not the same as adding words to your personal dictionary but works by taking the text that you pass it parses it into words and then records the relative frequencies of occurrence of each word with respect to the words around it If you use documents that represent your style of speaking this feature will improve dictation accuracy Currently only MS Word 97 2008 doc files MS Word 2007 docx files and Text ba files are recognized Choose Browse to locate your Word doc file then choose Begin Add From File to begin Tips amp Tricks This section contains shortcuts and tips for using the WSR Toolkit Using Add From File With Multiple Files Currently the Add From File feature only analyzes one file at a time It may become burdensome if you have many files you would like for it to analyze You can of course create one file from multiple files Within Microsoft Word here s how 1 Create a new blank document 2 Choose the Insert tab ALT N 3 In the Te
15. ick Browse to choose a specific application 2 When Say Enter a word or phrase that will trigger the macro 3 Insert The Text Enter the text as you wish it to appear when you say the trigger word Note You can navigate this form and enter text by voice Just say what you see Assuming Speech Recognition is active and listening You say When say The cursor moves to the When Say text box area Notice the letter W is underlined in When say That means you can press and hold down the ALT then press the w and the cursor will move to this location In fact anytime you see an underlined letter you can hold down the ALT key while pressing that letter and you ll move the cursor to that area Once the cursor is in this textbox you can say or type your trigger word That is the word you want the Speech Macros system to listen for You say my return address is the text my return address is appears in the When Say textbox If you re speaking to this program and not typing you may sometime encounter recognition errors for instance instead of my return address it might think you said something like my return add rest is No problem just say correct add rest and you ll be presented with choices that sound similar to what you ve just said All you have to do now is say the number that corresponds to the correct representation If it still misun
16. in from text other than the Rainbow Passage choose Clear then copy paste your selected text into the textbox area When you are ready to begin choose Begin Training Text Macros Command Macros Macro Editor Train From Text Add To Dictionary Transcription Add From File About Train From Text The purpose of Train From Text is to improve the acoustic model of your user profile We suggest you train from The Rainbow Passage below This passage developed by linguists contains a good sampling of phonemes individulal sounds found in the English language If you wish to train other text click or say Clear then cut and paste desired text into the text box area below When you are ready to begin training click or say Begin Training The Speech Recognition Voice Training Wizard will guide you through the process The Rainbow Passage When the sunlight strikes raindrops in the air they act as a prism and form a rainbow The rainbow is a division of white light into many beautiful colors These take the shape of a long round arch with its path high above and its two ends apparently beyond the horizon There is according to legend a boiling pot of gold at one end People look but no one ever finds it When a man looks for something beyond his reach his friends say he is looking for the pot of gold at the end of the rainbow Throughout the centuries people have explained the rainbow in vari
17. installation checks your machine for the Microsoft NET 3 5 Framework or greater and if it s not installed it will ask your permission to install After installation there will be three shortcuts on your desktop to the WSR Toolkit application the user s manual and the readme txt file Once the WSR Macro facility is installed and running you ll want to set macro security level to low as follows 1 Right click on the Windows Speech Recognition Macros icon in your system tray bottom right corner of your computer screen 2 Click Security then Set security Level Choose Low This is an acceptable option while you become familiar with the WSR Toolkit and develop your macros You can always come back to here and set the security level to high as well as digitally sign your macros once you gain more familiarity with the macro system You may decide it is desirable to leave the security level set to low Text Macros On this tab you ll be able to create macros that insert text into a document or other application Take a look at the screenshot for this tab and then we ll walk through the creation of a typical Text Macro Windows Speech Recognition Toolkit Tet ase O In The Context Of Gibai Al Applications When Say About Text Macros Allows you to create reusuable text Steps For Creating a Text Macro 1 In The Context Of Choose whether you want the macro to work in all applications global or cl
18. ous ways Some have accepted it as a miracle without physical explanation CE Add To Dictionary The purpose of Add To Dictionary tab is to help the speech recognition engine recognize difficult to understand words and phrases If you find certain words or phrases consistently being misunderstood you can add these to your personal speech dictionary as well as record a pronunciation This will increase the chances of accurate recognition for these particular words and phrases To use this feature copy and paste words and or phrases in the text box area Each word or phrase should be on a separate line for instance alpha beta zebras at the zoo etcetera and so forth When you are ready to begin adding these words to your personal dictionary click or say Add Words and you will be guided through the process by the Speech Dictionary Wizard where you will be given the option of recording a pronunciation for each word or phrase Important Note If you would like to edit change or delete words from your speech dictionary leave the list blank and say or click Add To Dictionary Here s a screenshot of the Add To Dictionary tab loaded with some of my troublesome words For instance the word coupon instead of saying coo pon prefer to say q pon The only way to get speech recognition to recognize this was to add the word to the dictionary with a recorded pronunciation This is important to no
19. owing things when you say copy web address 1 Move the cursor to the address bar 2 Copy the web address of the current website into the clipboard Now all you have to do is go to another application and say paste and it will put that address you just copied in the application This is nice when you d like to email a friend a website address for instance Here s how to do it On the Command Macros tab You say In the context of The dropdown box drops down revealing commonly used programs You say firefox Firefox exe is displayed in the In the Context Of text area You say alt The ALT checkbox will be checked You say numbers letters words The cursor moves to the Numbers Letters Words textbox You say d The letter d is placed in the Numbers Letters Words textbox The keyboard shortcut ALT d in Firefox will move the cursor to the address bar and highlight whatever address is present Notice the Keystroke Preview area This dynamically shows you the keystroke commands that are going to be built You say add The Keystroke Preview area clears and the Command Sequence Viewer displays a new line of text Now we need to insert a timing delay This is necessary because it s possible for the voice recognition system to emulate key presses faster than the application in this case Firefox can digest them So we need to insert a negligible
20. sert a timing delay 12 Keystroke Preview 11 Microsoft NET 3 5 Framework 4 Numbers Letters Words textbox 11 surf the net 9 trigger word 6 WSR 2
21. te It is highly recommended that you always record a pronunciation when adding words to the dictionary It will result in more accurate recognition than just adding words SS Add To Dictionary Transcription Add From File About Add To Dictionary The purpose of Add To Dictionary is to help the speech recognition engine with difficult to understand words and phrases ea find certain words or phrases consistently After copying pasting or manually adding your words click Add Words etcetera and so forth When you are ready to begin adding words to your personal dictionary click or say Add Words and you will be guided through the process by the Speech Dictionary Wizard where you will be given the option of recording a pronunciation for each word or phrase Note Ifyou would like to edit your speech dictionary just click or say Add Words with no words in the text box Transcription An important note on transcription There will be a noticeable increase in recognition accuracy if you use the Train From Text and Add From Text features before using transcription If you have a portable voice recorder this feature may be just what you re looking for With a portable voice recorder you can dictate notes memos and record thoughts and ideas while you re away from the computer Once you return just copy the sound file to your computer convert to a wave WAV file note some
22. u can send keys to a particular application It must be terminated with a lt sendKeys gt tag In Microsoft Word you may want to send the keys CTRL p to bring up the print dialog In your macro code it would look like this lt sendKeys gt CTRL p lt sendKeys gt You can find more on the syntax of the send keys here http blogs msdn com robch archive 2008 05 07 sending keys using custom wsrmacros aspx lt insertText gt lt run gt Like the other tags so far this one also needs to be properly terminated with a lt insertText gt tag Like the lt sendKeys gt tag this tag will insert any text between the opening and closing tag Unlike lt sendKeys gt you can have plenty of space in between For example lt insertText gt Line 1 Line 2 Line 3 was blank and this is line 4 lt insertText gt This tag allows you to run commands similar to executing them from the command line The tag lt run command C Program Files Mozilla Firefox Firefox exe params gt will run the Firefox browser Notice you can also pass the program parameters with the params attribute This would be useful if for instance you d like to start notepad and open a particular text file It would look something like this lt run command notepad exe params MyFile txt gt Note this tag is terminated differently than the others so far There isn t a lt run gt tag lt applsinForeground gt lt applsinForeground pro
23. w msspeech forum com http tech groups yahoo com group ms speech Here s the screenshot highlighting some key areas and text Windows Speech Recognition Toolkit a S ae So Cena Mario About Command Macros Allows you to create multiple step by step commands using common keyboard shortcuts In The Context Of Global All Applications When Say Simulate Keypress E ALT Command Key CTRL SHIFT Numbers Letters Words Keystroke Preview Take another quick look at the Command Key dropdown box This dropdown box contains every special key that can be used with Windows For example the F1 key Here s a screenshot with a partial listing of command keys displayed a Mm a n ecognit on Toolkit Text Macros Command Macros Macro Editor Train From Text Add To Dictionary Transcription Add File About Command Macros Allows you to create multiple step by step commands using common keyboard shortcuts 7 In The Context Of unid 5 on a CTRL Down Simulate Keypress ALT Command Key en err E SHIFT BS f i i A l oF Add a Program tc NUMLOCK E PGDN PGUP RIGHT This is how and where you d craft command sequences like ALT HOME SHIFT F10 etc The overall concept with command macros is simple If you know the keyboard shortcut key
24. xt ribbon box choose Object Text from File A 2 Signature Line e 8 Date 8 Time Drop Cap ed Object gt Text g4 Object t You can now select multiple files and they will be concatenated into one large file Save this file within MS Word You can now browse on the Add From File tab to this large file Increasing Speech Recognition Accurac Following is a recommended procedure for increasing speech recognition accuracy with the Toolkit Note it assumes you are using a new speech profile 1 Set up the microphone in Windows Vista you ll read the Peter dictates to his computer passage 2 Run WSR Toolkit and read the Rainbow Passage on the Train From Text tab 3 Select the Add From File tab and browse select one of your MS Word document or plain text file that represents your style of writing speaking Additional tips for increased recognition accuracy 1 Be diligent about correcting mistakes using correct that instead of simply retyping 2 Add words or phrases to your personal speech dictionary WSR Toolkit Add To Dictionary tab that are repeatedly misrecognized First try adding them without pronunciation then if it still gets it wrong delete and re add the word or phrase with pronunciation Index Command Key 15 Command Sequence Viewer Up Down amp Remove Buttons 13 context Global Context 7 copy web address 11 in
Download Pdf Manuals
Related Search