CoolSoft, LLC Making it FUN and EASY to use SPEECH with your COMPUTER!
User Manual for Speak-to-Mail 1.0
– Version 1.00.23
INTRODUCTION
Speak-to-Mail™ is a simple, low-cost email program
that is fun and easy to use. It lets you
send email by speech using your favorite email program and state of the art
speech recognition technology from Microsoft Corporation. Now, with Speak-to-Mail™, you can send emails
simply by speaking to your computer.
How does Speak-to-Mail™ work? Easy – it is a interface for writing an email with speech and passes your message to your email program where it is sent according to how your email program is configured to send email. Messages sent with Speak-to-Mail™ are kept in the sent folder of your default email client just like any other message you send. Speak-to-Mail™ comes in three editions to support Microsoft Outlook Express, Microsoft Outlook and Netscape 7.x.
Speak-to-Mail™ is user friendly and very easy to
use. It displays the contents of the
address book from your email program, and when you say the recipient’s name, it
places their email address in the desired field (TO, CC, BCC). You can even sort your contacts by first or
last name. You can then dictate the subject
and body, and insert attachments entirely by speech.
Note: CoolSoft, LLC acquired the rights to Speak-to-Mail 1.0 and other assets of Yellow Tiger, LLC in December 2003. CoolSoft is re-releasing Speak-to-Mail™ Version 1.00. 23. There are some important differences between Version 1.00.23 and the previous version 1.00.22 formerly released by Yellow Tiger, LLC.
·
The full version of Speak-to-Mail 1.00.23 no longer requires a registration
key, and can be obtained only by purchasing a license from www.coolsoftllc.com.
·
Speak-to-Mail™ now comes in a single install that includes the speech
engine and lets you choose which edition you want to install. There are three editions for Outlook Express,
Outlook and Netscape 7.x.
·
Editions for Netscape 4.x, 6.x, Eudora and AOL have been discontinued.
·
Bug fixes to some known issues identified in version 1.00.22 have been
made.
·
SpeakPad™ has been discontinued, and replaced by SpeakToText™ 2.0, a more
robust and feature packed speech program, which can be purchased at www.coolsoftllc.com.
·
Support for version 1.00.22 and all previous versions has
been discontinued.
·
The trial version of Speak-to-Mail™ remains at version 1.00.22. It does not include speech recognition, and
to use the trial version you must already have Microsoft SAPI speech
recognition installed on your computer.
·
The trial version of Speak-to-Mail™ will run only for 60 days and
registration keys to upgrade the trial version to the full version are no
longer available. To run the full
version of Speak-to-Mail™, the trial version must be uninstalled.
· The install of the full version gives you the option to uninstall of the trial version before continuing with the install.
Table of Contents
Setting Up the Microsoft SAPI 5.1 Speech Recognition
Engine
Technical Support on the Microsoft SAPI 5.1 Speech
Recognition Engine
Tips For Achieving the Best Possible Speech Recognition
Setting the Correct Default Email Program and Address Book
Replying to Emails from Your Email Program
Natural Language Modeling Technology
SpeakPad™ and the SpeakPad™ Button
SOLUTIONS FOR POOR RECOGNITION
GETTING STARTED
System Requirements
System requirements for Speak-to-Mail™ are:
|
Operating System: |
Windows 98se, Me, 2000 or XP (Home or
Professional) |
|
Memory: |
At least 128 Megs RAM, 256 or better
preferred. |
|
System Processor: |
A PC with a 500mhz or greater Pentium III,
Pentium IV or Pentium compatible processor.
|
|
Disk Space: |
100 megs free disk space |
|
Software: |
Microsoft SAPI 5.1 Speech Recognition
Engine.* At least one of the following email programs: Outlook, Outlook
Express or Netscape 7.x |
|
Hardware: |
A good quality, noise-cancellation
microphone. A good quality, SoundBlaster compatible sound card |
|
User: |
About 15 to 30 minutes of your time to
install Speak-to-Mail™, install the redistribution of the Microsoft SAPI 5.1
speech recognition engine, and train your recognition profile to recognize
your voice. |
*Available by redistribution with Speak-to-Mail™ or
directly from Microsoft.
Note: The trial version of Speak-to-Mail does not include the SAPI Speech
Recognition Engine. To use the trial
version of Speak-to-Mail, you must already have SAPI 5.0 (available with Office
XP), SAPI 5.1 (available with the full version of Speak-to-Mail™ 1.00.23,
SpeakToText™ 2.0 or any other speech recognition program that redistributes the
SAPI 5.1 speech recognition engine), or SAPI 6.0 (available with Office 2003).
Setting Up the Microsoft SAPI 5.1 Speech Recognition Engine
NOTE: If this is your first time using SAPI speech recognition, it’s important that you create a personal speech recognition profile, configure your microphone and train the speech recognition engine to recognize your voice. We cannot over emphasize the importance of these steps. This is done in “Speech Properties” which is part of the SAPI Speech Recognition Engine. “Speech Properties” can be accessed from the Control Panel, or by the Speech Properties menu item.
Start “Speech Properties”
Go to the control panel (StartàSettingsàControl Panel). Find and double
click the “Speech” icon.
The “Speech Properties” dialog box will appear.
Setup your Recognition Profile
You can use the “Default Speech Profile,” but we
recommend you create a speech profile with a unique name – especially if there
will be others using speech with this computer. To create a new profile, click
on “New…” and follow the prompts.
When have completed creating your new profile click “Next” to configure your microphone and train your profile (do not click “finish”). The next two sections are extremely important in achieving good speech recognition and the highest possible level of accuracy.
Setup your Microphone
Microphone quality:
In selecting the correct microphone headset for your
use you need to consider: 1. Sound in the environment (A noise canceling
microphone is required.). 2. Fit and style (Will you be comfortable?). Cost is
usually not a factor. A less expensive one may fit your needs better than the
more costly. There are many suppliers of high quality speech recognition
microphones. Please, check our website
to see which ones we recommend.
Configure Microphone:
You should establish a habit of configuring your
microphone in the same way each time you start a speech recognition session.
Use the “Configure Microphone” button on in “Speech Properties” as often as you
need.
Select Audio Input
The “Audio Input” should be set to your system’s
default audio device. If your system has more than one audio card you can use
this button to select a new audio input device. If you are not sure which is
the correct device, please see the user manual for your computer or contact
technical support for the manufacturer of your computer.
Train the Microsoft SAPI 5.1
Speech Recognition Engine
Training Overview
Although the Microsoft Speech Recognition Engine is
much better than previous speech engines, it still needs a sample of each users speech to correctly convert sound into written words.
Each speaker and each computer audio system have unique characteristics, which
must be sampled and used to adjust the engine. This process is called
“training” the engine.
The initial training of your profile will take about 15 minutes and each user must train their own profile. During this process you should speak in a normal tone of voice and insure that you have configured your microphone properly.
Training
To start training click on the “Train Profile”
button from the Speech Recognition tab and follow the instructions. When you
first establish your profile the training session is chosen for you. If this is
not the first time you have trained this profile, you will be given a choice of
literature to read aloud to your computer for training.
Select a title and click
on “Next>.” Read and follow the instructions to complete your training
session.
Note: If it has been sometime since the initial
training of this profile, we recommend that you use the “Introduction to
Microsoft Speech Recognition” session to familiarize yourself with how training
works.
If you want to take a break, click on the “Pause”
button. Training will be paused and the
“Pause” button will change to “Resume”.
When you are ready to start reading again, click the “Resume” button.
When Voice Training has finished processing, you
will have the option to select “Finish” or to choose “More Training…”
IMPORTANT: To keep improving your accuracy, you should come back to the Speech Properties and do additional training. The more you train and use the speech engine, the more your accuracy will improve. In testing, we found a substantial improvement in accuracy between the second and third training. After five sessions, the accuracy was much greater. We suggest you take advantage of the opportunity to read all of the available selections to your computer.
Other Settings under the
Speech Recognition Tab
Language
The language must be set to “Microsoft English
Recognizer v5.1” in order for Speak-to-Mail™ to work properly. Do not change the speech recognition engine,
except to select “Microsoft English Recognizer v5.1” if that is not the
default. Do not change the “Settings” to
the right of the drop-down box either.
If you change these, Speak-to-Mail™ will not work properly.
Recognition Profiles
New
Click on the “New” button to create a new
recognition profile.
Delete
Click on the “Delete” button to delete the profile
that is highlighted.
Selecting Profile
Put a check mark next to the profile you want to
use. If more than one user will be using
Speak-to-Mail™ on this computer, each user should go to Speech Properties and
check to make certain his or her profile is selected each time before using the
program.
Settings
We do not recommend that you change the “Settings…”
for your profile. In most cases, the
default settings will give you the best recognition.
If, after several “trainings” and insuring your microphone
is configured properly you still difficulty with the accuracy of your
recognition, you may want to adjust the “Accuracy vs. Recognition Response
Time” setting or the “Pronunciation Sensitivity” to see if this will give you
improved accuracy.
Leave the “Background Adaptation” enabled (checked)
as this will allows the speech engine to continue
training as you use it. Over time and
use, the background adaptation feature will help improve your level of
accuracy.
If you have more questions about these settings, we suggest you contact Microsoft technical support.
Text To Speech Tab
The “Text To Speech” tab in the Speech Properties
dialog box is where you can select the computer’s output voice.
Voice Selection
Use the drop-down list under “Voice Selection” to
choose a voice, and you will hear the selected voice read a sample
sentence.
Preview Voice
Click on this button to hear a sample of the voice
you have selected.
Settings
We do not recommend that you change the settings
under the “Settings” button.
Voice Speed
You can adjust the speed of the voice and test it
again by clicking on “Preview Voice” button until you find a speed that you
like.
“Other” Tab
Help
Microsoft provides complete details about speech
recognition in the help section. To
learn more about the using the Microsoft speech recognition engine, go to the
Speech Properties dialog box, choose the “Other” tab and select “Help.” You can also find resources in the speech
section of the Microsoft website:
http://www.microsoft.com/speech/
Technical Support on the Microsoft SAPI 5.1 Speech Recognition Engine
The redistribution of the SAPI 5.1 speech
recognition engine is a Microsoft product that we do not support. The Microsoft speech recognition is designed
to enable you to use speech recognition with certain Microsoft programs and
other third party software, like Speak-to-Mail™, and in most cases should not
require technical support. However, if
you do need technical support on issues with the Microsoft SAPI 5.1 speech
recognition engine go to the speech section of the Microsoft website – http://www.microsoft.com/speech. Or
Contact Microsoft technical support.
USING SPEAK-TO-MAIL™
Tips For Achieving the Best Possible Speech Recognition
While speech recognition technology has advanced
dramatically over the past few years, the technology is still far from
perfect. Even the best and most
expensive speech recognition systems rarely achieve accuracy greater than
98%.
Although the Microsoft speech engine performs well,
it is not reasonable to expect accuracy greater than 90% to 95%. This is not a flaw in Speak-to-Mail™ or the
SAPI 5.1 speech engine, but rather is inherent to the current state of speech
technology. We believe that the
Microsoft SAPI 5.1 speech engine delivers comparable performance to that of its
counterparts that are available commercially.
Here are some tips on getting the best possible speech recognition:
Microphone – Using a high quality, noise-canceling microphone is very important. Making sure your microphone is setup properly in the Speech Properties is essential.
Training -
Training the engine to recognize your voice
at least several times when you first start using it will greatly improve your
recognition.
Training
Words – If you consistently get incorrect recognition of
certain words, using the “Train Word” function in Speak-to-Mail™ will greatly
enhance recognition of those words.
Use - The more you use speech with your computer, the more your recognition will improve.
Environment - Being in a quiet room helps.
Hardware - Having a faster processor and ample memory (256k sufficient, 512k or more is desirable) allows the speech engine to process speech more quickly and efficiently. Having a high quality, SoundBlaster compatible, audio input is also an important factor to good recognition.
Commitment - Uses must invest the necessary time at the beginning to set up and train their personal speech recognition profile at least several times. We cannot over emphasize the importance of training your recognition profile.
Speak
clearly - When using Speak-to-Mail™, speak clearly and
into the microphone. Make certain your
microphone position is consistent both in training and in use.
Following these tips will
help you get the most out of speech recognition, and will help make the process
more rewarding.
Setting the Correct Default Email Program and Address Book
In order for Speak-to-Mail™ to function properly,
the email program for the edition of Speak-to-Mail™ that you are using must be
set as the default email program, or said differently, you need to use the
edition of Speak-to-Mail that corresponds with your default email program.
In most cases, you will not be required to changes
theses settings but if you have more than one email program you will need to
verify theses settings.
For example: If you are using the Outlook Express
edition of Speak-to-Mail™, and you also have Outlook installed on your
computer, then Speak-to-Mail™ will use the default email program and address
book.
To set your email program and address book as
defaults, you need to do the following:
Go to Internet Options in the Control Panel: (Start,
à Settings, à Control Panel)
Double click on the Internet Options icon. Then select the “Programs” tab.
From the “Programs” tab, in the “E-mail:” section
drop-down list, select the email program you want to set as default.
If you have both Outlook and Outlook Express
installed on your computer, and you are using the Outlook Express edition of
Speak-to-Mail™, then from the drop-down list in “Contact list, select the
address book you want to use as the default.”
Click “Apply” or “Ok” to apply the settings.
Starting Speak-to-Mail™
To start Speak-to-Mail™, either click the
Speak-to-Mail™ icon on your desktop, or use the program menu (StartàProgramsàSpeak-to-Mailà“Speak-to-Mail for
(Edition)”.
Note: The trial version of Speak-to-Mail™ is
version 1.00.22, and will display the startup splash screen, which enables a
60-day trial period. The trial version
cannot be registered or upgraded to the full version, and must be uninstalled
before you can install the full version 1.00.23. When you purchase the full version of
Speak-to-Mail™, the install will give you the option to uninstall the trial
version before continuing. Aside from
these differences and some bug fixes, the trial version 1.00.22 and the full
version 1.00.23 are substantially the same.
Using The Default Email
Program
As Speak-to-Mail™ is opening; the program reads the
address book from your email program for the edition of Speak-to-Mail™ that you
are using.
You must have the correct email program and address
book set as the default under Internet Options.
See “Setting
the Correct Default Email Program and Address Book” above for specific
details.
All emails must have a recipient and subject in
order for Speak-to-Mail™ to send the message; and your messages will be sent
according to the settings of your email program.
If your email program is not configured to send
messages immediately, you will have to go to your email program to send your
emails as you normally do. If your email program is configured to send messages
immediately, the message will be sent and a copy placed in the sent folder of
your email program.
Sending an email will clear all the message fields in Speak-to-Mail™, and the program will be ready to accept your next message.
Outlook and Outlook Express
When you start Speak-to-Mail™ for Outlook and
Outlook Express the email program is automatically opened and must remain open
for the program to function properly. See “Email Security” below.
Netscape 7.x
With the Netscape 7.x edition, the browser is opened
when you start Speak-to-Mail™. When sending an email, Netscape Messenger and a
compose window are opened and your message is placed in the Compose
Window. You can send the message now or
later, and the message will be placed in your default Sent Messages or Unsent
Messages folder depend on your selection.
Note that with Netscape 7.x, you MUST be connected to the Internet to send your message from Speak-to-Mail™. If you are not connected to the Internet, then depending on your system configuration, you may encounter errors that may cause your message not to be sent and prevent you from sending additional messages.
EMAIL Security
There are several security features that prevent
viruses from sending emails to people in your contact list and prevent the
sending of unauthorized emails. In order for Speak-to-Mail™ to function
correctly, you must allow it to access you address book and send emails. In this section we have provided information,
which will assist you in maintaining this security while using Speak-to-Mail™.
Outlook Express
If you are using Microsoft Outlook Express from an
installation of Windows 2000 or XP, and if you have the latest Windows security
updates, the Microsoft security service dialog box may appear when sending
emails.
If you have this feature, your email will not be
sent and the message fields will not clear until you click on “Send”. If you click on “Do Not Send”, you will not
be able to send your email.
Outlook
If you are using Microsoft Outlook 2000, XP or 2003,
from an installation of Windows 2000 or XP, and if you have the latest Windows
security updates, the Microsoft security service dialog box may appear.
Check the box labeled “Allow access for”, use the default of “1 minute” and click on “Yes”. This will allow Speak-to-Mail™ to access the address book and read your contact list. Speak-to-Mail™ will not finish loading until you have made a selection.
When sending an email with Outlook you may receive
this message.
This feature is similar to the security message
displayed in the Section – “Starting Speak-to-Mail™,” except in this instance,
instead of asking for permission to access the address book, you are being
asked if you want to allow Outlook or Outlook Express to send an email. Click on “Yes” to allow your email to be
sent.
More on Email Security
If you do not see one of the security dialog boxes
described above, you either do not have the latest security updates or the
security dialog box is minimized in the toolbar at the bottom of your
desktop. To see if it is minimized, look
in the toolbar at the bottom of your desktop (see below). A left mouse click will bring it to the
foreground and display the security dialog box as shown above.
or 
Outlook Outlook Express
We are not currently aware of security features that
protect the address book in Outlook Express or Netscape, or prevent
other programs from sending emails in Netscape but it is possible that
such features could be added in future updates.
If these security features become available, you will need to follow similar procedures to those described above. We will post updates to this manual and instructions on our website as we learn about these updates.
Speak-to-Mail™ Interface
When Speak-to-Mail™ is finished loading; the
Speak-to-Mail™ interface will display the names from your contact list on the
left and an email template on the right.
Command buttons, a dialog box that lets you see what the speech engine
is recognizing, and a black box that differentiates dictation and natural
language are all located in the lower part of the screen.
Three different styles of buttons are used in the Speak-to-Mail™ interface and its sub-screens:
YELLOW - The buttons with yellow text are where you email addresses and text are entered and activated by speech commands (i.e. “select (NAME)” to place a contact’s name in the “To:” field.) They do not perform any task when you click on them.
GREEN - The buttons with green text are function buttons that can be activated both by speech commands and by mouse clicks.
GRAY - Gray buttons, like the buttons typically used in Windows applications, are used in Speak-to-Mail™ sub-screens where speech recognition is not active. Examples are the attachments and signature file settings dialog boxes. A gray button indicates that speech recognition will not work with that button, and that you must use your mouse to click on it.
Important Notes
IMPORTANT ITEMS TO NOTE BEFORE YOU START USING Speak-to-Mail™
Microphone On-Off
Speak-to-Mail™ opens with the microphone in the off
position to prevent Speak-to-Mail™ from picking up background noise or
recognizing your speech before you are ready to dictate.
To start using Speak-to-Mail™, you must activate the
microphone by saying “Wake Up” or “Microphone On.” You can also click on the microphone icon in
the lower part of the screen to activate the microphone.
When you return to Speak-to-Mail™ after going into
“About” or “Update Signature”, the microphone will be turned off. You need to turn the microphone on
again.
Turn the microphone off when you are done sending
messages. You can leave Speak-to-Mail™,
running in the background with the microphone turned off. But, it is important to turn the microphone
off before Speak-to-Mail™ goes into the background. Otherwise, the program will continue to
recognize any background noise or speech.
To turn off the microphone, say, "Go to sleep” or “Microphone off.”
Speech Commands
Speech commands are used in Speak-to-Mail™ to
activate buttons on the screen in place of clicking on them with your mouse.
Speech Commands are also used within Speak-to-Mail™ for navigating the screen
and placing email addresses in the desired fields. These do not appear on the screen as buttons.
Dictation Commands
Both Speak-to-Mail™ and the Microsoft speech
recognition engine use dictation commands.
An example of a dictation command used by Speak-to-Mail™ is, “scratch
that,” which will erase the last word you said.
Another example of a dictation command is the “forcenum” command. Saying “forcenum” with a number between zero
and twenty will cause that number to by typed as a
numeral instead of being spelled out as a word, which is the default for the
speech engine.
In speech recognition, commands work best if they consist of two or more words that would not normally be said in a sentence together. Because of this, commands (for example “send email”) require you to say, “click” before the command, so to send an email you have to say, “click send email.”
IT IS IMPORTANT THAT YOU DO NOT PAUSE BETWEEN WORDS
IN A COMMAND, OR THE SPEECH RECOGNITION ENGINE MAY RECOGNIZE THEM AS TEXT
INSTEAD OF A COMMAND. If this happens,
simply repeat your command without pausing between words.
To help you remember what you have to say, we have
included Tool Tips for each button in Speak-to-Mail™. When you place the arrow over a button, the
tool tips show you what you need to say to activate the command for that
button.
Since Dictation Commands do not have buttons, those commands do not have Tool Tips. When beginning to use Speak-to-Mail™, we recommend that you keep a copy of the Speak-to-Mail™ documentation handy to help you with dictation commands.
Address Book
Speak-to-Mail™ uses the address book of the email
program for the edition
of Speak-to-Mail™ you are
using. In order for Speak-to-Mail™ to
work properly, you need to have your contact’s first name, last name and email
address entered into the address book of the email program you are using with
Speak-to-Mail™.
While you can type any email address into any of the
address fields in Speak-to-Mail™, it is best to have each of your contacts
entered into the address book so the contact’s name will appear in the
Speak-to-Mail™ list box on the left. You
will then be able to speak a contact’s name and their email address will be
placed in the field where you want it to go.
Starting Speak-to-Mail™,
if your contact list is empty, may generate errors. You will still be able to continue and use
Speak-to-Mail™, but you will have to use the keyboard to enter email addresses.
If you have phone numbers in your contact list, Speak-to-Mail™ can also find phone numbers for your contacts.
Resetting the names list
Each time start Speak-to-Mail™, the program reads
the address book from the email client you are using and builds the list box
you see on the left. If you add a
contact into your email client or make any other changes after you have opened
Speak-to-Mail™, any new contacts or changes will not appear in the list box
until you close and restart Speak-to-Mail™.
All the names in your address book must be unique.
If Speak-to-Mail™ finds more than one entry for a
name when it is reading your address book, it will load only the first
occurrence of that name. For example, if
you have two John Doe’s in your address book, only the first one will appear in
the Speak-to-Mail™ list box.
If you have duplicate names in your address book, you need to make them all unique so none of the contacts are missing from the Speak-to-Mail™ list box. (Examples: [John Doe, John Doe Home], [J Doe, John Doe], [John Doe Home, John Doe Work], [John J. Doe, John Doe], [John Doe, Sr., John Doe, Jr.].). You can use these examples or come up with combinations of your own, provided there are no duplicate names.
The names in your address book must contain letters only.
Speak-to-Mail™ cannot recognize contact names that
contain characters other than letters.
When Speak-to-Mail™ loads your contact list it removes all extra
characters and displays only the letters.
For example, an entry like “John Doe (work)” will appear as “John Doe
work” and has to be spoken that way.
Getting ahead of the speech
engine
If you are dictating an email and the recognition
starts to deteriorate, sometimes pausing for a short while to allow the speech
engine to catch up will solve the problem. If pausing does not improve
recognition, then see Section “Known Issues” for other possible solutions.
Recognition Accuracy
The accuracy of speech recognition should continue
to improve each time you train the speech recognition engine. Time invested in training is well worth the
return in improved accuracy. Finally, it is important that you are in a quiet
environment with minimal background noise, and that you always speak clearly
into the microphone in order to achieve the most accurate possible recognition.
Speech Modes
Speak-to-Mail™ uses three speech modes:
Ø
Command & Control
Ø
Dictation
Ø
Natural Language Modeling
These modes, for the most part, are transparent to
the user and are discussed briefly below to give you an overview of how they
function. Try experimenting with different ways of sending an email to see what
works best for you.
Command & Control
Command & Control is used to navigate the fields
and perform functions.
For example, to navigate fields, you can say “to” and the cursor will appear in the “To” line. Say “select (contact name)” and the contact’s email address will appear in the “TO” field. Say “subject” and the cursor will appear in the “Subject” field. Say “body” and the cursor will go to the “Body” of the message.
Command Accuracy
In speech recognition technology, function commands
are best recognized if they consist of two or more words that are not likely to
be spoken in natural speech. Examples of
function commands are “click send email” and Speak-to-Mail™ will send your
message to the outbox of your email program. Say “Clear Message” to clear a message and
start over. Say “Click Close” to exit
Speak-to-Mail™.
Dictation is used to freely dictate your email
subject and text message. When you are
in the “Subject” field or in the “Body” field, the dictation mode activates
automatically.
If you want to say a command, you need to pause dictation briefly, say the command and pause
again. For example, to send an email
using dictation, you could say the following:
“Select John Doe”
“Select Jane Doe”
“Copy to Jane Smith”
“Blind Copy to John Smith”
“Subject [pause] Next Tuesday’s meeting”
“Body”
In the example above, the email addresses for John
Doe and Jane Doe will appear in the “To” field, Jane Smith in the “CC” field
and John Smith in the “BCC” field. “Next
Tuesday’s meeting” will appear in the “Subject” field, and the cursor will be
at the beginning of the body ready to accept dictation.
Training Words
If the speech recognition engine consistently has a
problem recognizing your pronunciation of certain words, you can train the
Microsoft speech recognition engine to recognize how you say that word. To train a word, you must be in
Speak-to-Mail™. Say, “click train word”
or click on the “train word” button with your mouse, and the Train Word dialog
box will open.
Type in the correct
spelling of the word you want to train in the “Word” box, then click on the
“Record pronunciation” button and say the word.
After the computer recognizes your pronunciation of the word, it will be
added to a custom dictionary for your recognition profile. This will improve the speech engine’s
recognition of that word when you say it.
You can train words from both Speak-to-Mail™ and the
trained words will be saved to a library within your recognition profile. These trained words can be recognized from
both programs, but remember, training is particular to each individual
recognition profile. Each person that
uses speech recognition on your computer needs to have their own recognition
profile and needs to train their own library of words.
If there are words that are particular to your usage
or occupation, it is a good idea to train these words. We suggest you train the “forcenum”
command. If you want to have numbers
zero to twenty appear as numerals, then you will need to say “forcenum
(numeral).” For example, “forcenum two”
will return the number 2. The SAPI 5.1
speech recognition engine is programmed to spell out numbers from zero to
twenty, and we have found, in testing, that with most people it has difficulty
recognizing this command, unless it has been trained first.
To further demonstrate the power of training words,
try training the following word:
Speak-to-Mail(tm)
Then go into Speak-to-Mail™ and try saying “speak to
mail” to see the result. The phrase
“speak to mail” should now be recognized as “Speak-to-Mail(tm)”.
NOTE:
Custom training of words is a function controlled by the SAPI speech
recognition engine outside of Speak-to-Mail™. You have to use the mouse and
keyboard while you are in this dialog box.
Speech recognition resumes when you close the Train Word dialog box to
return to Speak-to-Mail™.
Clearing the Speak-to-Mail™ Speech Buffer
At times, the accuracy of
your speech recognition may appear to stall when there is too much speech in
the buffer. The delay results while the
speech recognition engine is processing all the data it has received. This delay can be caused by talking too much
too fast, or when the computer picks up background noise or conversations of
other people. This should not occur frequently, especially with faster
computers that have a lot of RAM.
Usually, if you pause for a few moments, the speech engine will catch up
and you may not need to clear the buffer.
If you want to clear the speech buffer immediately,
you can by pressing the escape “Esc” key.
You will be asked to confirm that you want to clear the buffer. Click “ok” and all unrecognized speech within
the Speak-to-Mail™ buffer will be cleared and the program will be ready start
dictation from scratch.
NOTE:
Clearing the buffer in Speak-to-Mail™ will not clear any unrecognized
text that is already being processed by the speech recognition engine. You may still need to wait a few moments
while the speech recognition engine completes the recognition already in
process.
NOTE: If Speak-to-Mail™ is open and you try to use
the “Esc” key in another application, this can cause an error in
Speak-to-Mail™. If you are going to use
the “Esc” key in another application, it is best to close Speak-to-Mail™ first.
Using the Mouse and Keyboard
If you are having trouble with recognition, or do
not have your microphone handy, you can use the mouse and keyboard to send an
email with Speak-to-Mail™.
Use the mouse to navigate to the fields and to click
buttons on the Speak-to-Mail™ screen.
Right click on a contact’s name and a menu pops up
that enables you to place that contacts name in the desired field.
Use the keyboard to type in email addresses that are
not in your contact list, or are not being recognized.
Use the keyboard to type the subject and text of
your message.
In the “Subject” or “Body” fields, you can use the
mouse to highlight the desired text, and then you can use the right mouse
button to cut, copy or paste text.
Remember, if you are cutting or copying text, your selection must be highlighted first. To paste text, you must use the mouse to place the cursor in the location where you want to paste the text.
Additional Mouse Functions:
When Speak-to-Mail™ opens, the default setting for a
single left mouse click on a contact’s name places the contact’s email address
in the “To:” field.
Whenever you use speech to place a contact’s email
address in another field or ask for a phone number, the setting for a single
left mouse click will change to that command.
For example, saying “copy to (contact name)” will
change the setting for a left mouse click to the “Cc:” field, and after that,
whenever you left click on a contacts name their email address will be placed
in the “Cc:” field.
The same applies to the “Bcc:” field, requesting
phone numbers, or returning the setting to the “To:” field. Using one of those speech commands will
change the setting for a single left mouse click on a contact’s name to the
last email command (or phone number command) that was executed by speech.
Speak-to-Mail™ Commands
This section gives an overview of all commands that can be used with Speak-to-Mail™.
Tool Tips
To help you with commands, “Tool Tips” are included
with each button that is speech enabled.
While using Speak-to-Mail™, if you are not sure what to say, or have
forgotten place the arrow over a button and it will tell you what you can say
to activate the command function for that button.
Control Commands
Speak-to-Mail™ uses spoken commands (Command &
Control) to activate the functions of buttons on the Speak-to-Mail™ screen and
place email addresses in desired fields.
You can either say the command or click on the button.
In order to help the speech recognition engine
differentiate between dictation and control commands, and to prevent text words
from being recognized as commands, most commands generally consist of two or
more words that would not normally be used together in a sentence. For example, to close Speak-to-Mail™ you have
to say, “click close.”
Combining the words “click” and “close” minimizes the likelihood that
that the program will be closed by accident.
In order to make commands work best, do not pause in
the middle of a command. For example, if
the command is “click close,” do not pause between the words “click” and
“close.” A pause may cause the command to be recognized as text
If you are dictating text and you want to say a command, pause briefly so the speech engine knows that the next thing you say may be a command. You do not need to pause before or during a Dictation Command; just speak it as part of your sentence. (See Section – “Dictation Commands”.)
Microphone Commands
To turn on the microphone, say “Wake Up” or
“Microphone On”.
To turn off the microphone, say “Go to Sleep” or
“Microphone Off”.
You can also click on the microphone icon to turn
the microphone on and off.
Microphone Off Microphone On
Email Commands
Screen Navigation
To move the cursor to a field:
Desired Field Commands
(what to say)
To “To”
CC “CC”
“Copy”
BCC “BCC”
“Blind” “Blind Copy”
Subject “Subject”
“Email Subject”
Body “Body” “Text Message”
“Text” “Email Text” “Email Body” “Email Message”
You can also navigate the fields manually with the
mouse.
NOTE: There is really no need to navigate to the “TO,” “CC,” or “BCC” fields unless you plan to type an email address manually, however, we have included these commands for your convenience.
Placing a Contact’s Email Address in an Address Field
To place a contact’s email address in one of the
address fields, say the following where “NAME” is the name of the contact you
want to select. This is similar to
navigating the address fields as above, but you are saying the name of the
field and the contact’s name.
Desired Field What
to Say
To “Select NAME” “Send to NAME”
CC “Copy to NAME” “Copy NAME”
BCC “Blind to NAME” “Blind Copy NAME”
Left clicking on a contact’s name in the list box
will place the contact’s email address in the “TO” field or in the last field
that was used with a speech command.
Right clicking on a contact’s name will also allow
you to use the mouse to place the contact’s email address in the desired field
(see Section “Using
the Mouse and Keyboard”).
NOTE: In
order to make sure you are always using the most up-to-date contact
information, each time you start Speak-to-Mail™; the program imports a fresh
copy of the contact list from your email program for the edition of
Speak-to-Mail™ you are using.
Speak-to-Mail™ is cannot write contact information
back to your email program. If you want
to do that, you must add or change the contact information in your email
program. The new/modified contact will
not appear in Speak-to-Mail™ until the next time you start Speak-to-Mail™.
If any of your contacts do not have an email address
in your email program’s address book, Speak-to-Mail™ will tell you that your
<Contact Name> does not have an email address. To temporarily solve this problem, you can
type in your contact’s email address in the field where you want to place it.
If you are typing more than one email address in a field, you must separate them by a semi-colon “;” and a space. To permanently solve this problem, you need to enter the first and last name and email address of all your contacts into the contact list in your email program.
Clearing Fields
If you make a mistake and want to clear a field, say
the following:
Desired Field Commands (what to say)
To “Clear To” “Clear Send To”
CC “Clear
CC” “Clear Copy” “Clear Copy To”
BCC “Clear
BCC” “Clear Blind” “Clear Blind Copy” “Clear Blind Copy To”
Subject “Clear
Subject”
Attachment “Clear
Attachment”
Selected Text “Delete
Selection”
Body “Clear Text” “Clear Message” “Clear Body”
A dialog box appears asking you if you are sure you
want to clear the message. You have to
use your mouse to choose “Yes” or “No.”
Once a message is cleared, there is no way to recover it
All Fields “Clear Email” “Clear All”
A dialog box appears asking you if you are sure you
want to clear the email. You have to use
your mouse to choose “Yes” or “No.” Once
a message is cleared, there is no way to recover it.
Sorting Contacts
When Speak-to-Mail™ first opens, your contact list
is sorted by first name so you can say a contact’s first and last name as you
would in natural speech. You can
alternate sorting between first and last name simply by saying, “click sort.” You can also use your mouse to click on the
“Sort Contacts by First/Last Name” button.
Note that when your contact list is sorted by last
name, in order to select a name you have to say the contact’s last name, then
his or her first name. For example to
place John Doe’s email address in the “TO” field:
If contacts are sorted by first name, you have to
say “select John Doe.”
If contacts are sorted by last name, you have to
say, “select Doe John.”
Scrolling Up and Down the Contact List
To scroll down the contact list, say “contacts page
down.”
To scroll up the contact list, say “contacts page
up.”
You can also scroll up and down the contact list by using your mouse on the arrows to the right side of the contact listbox.
Finding a Contact’s Phone Number
If your contact list
includes phone numbers, Speak-to-Mail™ can retrieve a contact’s phone number
for you when you say “Phone Number (the contact’s name).”
The Home, Business and Mobile phone numbers will be
displayed in a text box on your screen, and the phone numbers will be read
aloud.
If your contact does not have any phone numbers, you
will get a message that your contact doesn’t have any phone numbers.
Note: In version 1.0, phone numbers only work in Outlook and Eudora we are working to add this feature to future releases.
Attachments
To open the Attachment Window, say “Click
Attachment.”
Once the Attachment Window is active, you have to
manually select the files you want to attach and click the “Save” or “Cancel”
button to return to the Speak-to-Mail™ main screen. Once you return to the main
screen, speech recognition is again active.
You can also click on the “Attachment” button with
your mouse to open the Attachment Window.
The Attachment dialog box in Speak-to-Mail™
functions the same way as it does in any other email program. You cannot use speech while placing an
attachment, but after selecting your attachment or closing the Attachment
dialog box, you will return to Speak-to-Mail™, where speech recognition is
again fully active.
If you have more than one attachment, they have to be entered one at a time. We recommend that you zip multiple attachments into one file.
Sending Your Email
To send your email, say “Click Send Email” and your
email will be sent. Sending an email will clear all the message fields in
Speak-to-Mail™, and the program will be ready to accept your next message.
Closing Speak-to-Mail™
To close Speak-to-Mail™, say “Click Close,” or click
on the “Close” button. Speak-to-Mail™
will display a dialog box asking you if you are sure you want to exit
Speak-to-Mail™. Click on “Yes” to exit Speak-to-Mail™, or on “No” if you do not want to exit. This confirmation feature prevents you from
losing your email by accidentally clicking on the close button. Note that the default for this box is set to
“Yes” so you can also press “enter” on your keyboard to exit the program.
When you close Speak-to-Mail™, if Speak-to-Mail™ opened your email program, Speak-to-Mail™ will also close it. If it was already open, then it will remain open.
Replying to Emails from Your Email Program
Many people like to reply to their messages instead
of originating a new message. The
process for doing this with Speak-to-Mail™ is simple and easy. Just use Speak-to-Mail™ to dictate your
reply. When you are done dictating, copy
your message to the clipboard, then go to your email program. Start a reply to your message, paste your
dictated text into the message body in your email program, and send your reply.
Speak-to-Mail™ Settings
Embedding a Signature File in Your Emails
Speak-to-Mail™ lets you embed a text signature at
the end of each email. You can do this
globally for all of your emails, or with individual emails.
Before you can embed a
signature in your email, you must first setup a signature file. To do this, say “click update signature” (or
click on the “Update Signature” button) and the Settings screen will open
Speech recognition is not active in this dialog
box. Use the keyboard to enter your
signature file as you would like it to appear and click on “Save.”
You can choose to have your signature in every email
by checking the “Enable Signature in every email” checkbox. The default setting is unchecked.
Click on “Save” to save your settings and return to
Speak-to-Mail™. Click on “Close” to
discard your changes. Remember to turn
the microphone on again after updating the signature settings.
If you want to individually select which emails will
have your signature, leave the “Enable signature in every email” box
unchecked. When you want to include your
signature in an email, say, “click insert signature” and your signature will be
added to that email only.
If you want to change your signature later, return to the update Signature”.
About Speak-to-Mail™
To open the “About” dialog box, say, “click about”
or click the “About” button with your mouse.
The About dialog box displays the user information,
registration key, and the version of the product.
You can also access the signature and greeting settings by clicking the “Settings” button in the About screen.
Speak-to-Mail™ Signature Message
Speak-to-Mail™ includes its own signature message at
the end of each email. This feature can
be disabled in the full version, but cannot be disabled in the trial
version. To disable/enable the
Speak-to-Mail™ signature message, say, “click Update Signature” or click on the
“Update Signature” button. From the settings
dialog box, uncheck or check the “Enable Speak-to-Mail banner” box according to
your preference. In the full version,
you can return to the settings dialog box whenever you want to enable or
disable the banner message.
If you like using Speak-to-Mail™ and you think it’s a great program like we do, we hope you will leave the banner enabled so that your friends and contacts can learn about it.
Speak-to-Mail™ Startup Greeting
When you are using Speak-to-Mail™ in trial mode, the
startup splash screen includes a textbox where you can type your name. If you leave this blank, the program will
greet you with “Welcome to Speak-to-Mail.”
If you enter your name in this box, then the program
will greet you with “Welcome (your Name).”
You can disable the greeting or change the name from
the signature settings dialog box. Just
say “Click Update Signature” or click on the Update Signature button to open
the signature dialog box and check or uncheck the greeting checkbox according
to your preference. You can also change
the name or enter a different name here.
Dictation Commands
When you enter the “Subject” or “Body” fields of
your email, Speak-to-Mail™ is ready for dictation, and you will receive a
message in the lower center part of the Speak-to-Mail™ screen.
The following sections give details on dictation commands that are used by Speak-to-Mail™ and the Microsoft SAPI 5.1 speech recognition engine.
Selecting Sentences and Paragraphs
To select the last sentence in a text message, say,
“select last sentence;” and the text up to the last period “.” will be
highlighted.
To select the last paragraph in a text message, say,
“select last paragraph;” and the last paragraph will be highlighted.
To select all the text, say, “select all,” and all the text in your
message will be highlighted.
You can also select text using the mouse by placing
the cursor at the beginning or end of the text you want to select and dragging
the mouse.
Once a selection is highlighted, if you say, “delete selection,” the selected text will be deleted and you can continue your dictation. Or you can simply dictate new text and the highlighted text will be overwritten.
Cut, Copy and Paste Commands:
You can copy all of the text to the clipboard by
saying “copy to clipboard.”
You can cut and copy selected text to the clipboard
by highlighting the text with the selection commands above (or with the mouse)
and using the right mouse key to cut or copy selected text. You can paste text
from the clipboard by placing the cursor where you want to insert the text and
using the right mouse key to paste text from the clipboard.
Important: If
you say something other than a command while text is highlighted, what you say
will replace the highlighted text and you will not be able to recover it. If you are going to talk while text is
highlighted, or if there is a lot of background noise, we suggest that you not
leave the text highlighted for a long period of time, or that you turn the
microphone off.
Correcting Text
As noted throughout this manual, today’s speech
recognition technology, although the best it has ever been, is not 100%
accurate. Even though you may be
speaking clearly from a quiet room, and are using a well-trained recognition
profile, and a high quality microphone and sound card, you cannot expect accuracy
to exceed 90% to 95%. You can use the
mouse and keyboard to correct recognition errors in your text.
To erase the last word you said, say “scratch that,” or “no no.” Each time you say this, the last word in the
text will be deleted.
Say “scratch last” to erase the last string of text
that was recognized. NOTE: Scratch last can only be used once. After you use the “scratch last” command, you
have to say something new before the command can be used again.
To delete selected (highlighted) text, say “delete
selection.”
Use the “copy to clipboard,” “cut to clipboard” or
“paste from clipboard” commands to copy, cut or paste highlighted text to and
from other applications. You can also
use the right mouse button while in the “Subject” or “Body” fields of Speak-to-Mail™
to “cut” or “copy” highlighted text so it can be copied into other
applications, or to “paste” text that has been cut or copied from other
applications. See - “Using the Mouse and
Keyboard” above.
To insert text before a specific word, place the
cursor before that word and begin speaking or use the keyboard.
To insert text after a specific word, place the cursor after that word and begin speaking or use the keyboard.
Moving Within Dictated Text
Command – What To Say: RESULT:
Go to Bottom move the cursor to the end of
your document.
Go to Top move the cursor to the beginning of your document.
Formatting Text
Command – What To Say: RESULT:
New
Paragraph Presses enter key twice.
New Line Presses enter key once.
Backspace Presses
the backspace key once.
Space Bar Presses space bar once.
Cap Text Capitalize the first
letter of each word in the selected (highlighted) text.
Cap All Text Capitalize all the letters
the selected (highlighted) text.
Bold Text Bolds the selected
(highlighted) text.
Italicize
Text Italicizes the selected
(highlighted) text.
Underline Text Underlines the selected (highlighted) text.
Numbers and Fractions
Number: What To Say:
Three three (the speech engine default for single
numerals between zero and twenty always typed as text by the Microsoft speech
engine)
“3” Forcenum three – will return the
numeral 3. You may need to train the
word “forcenum” before the speech engine will recognize it. See Section “Training Words.”
“34” thirty
four, or three four
“153”
”6091” six
eight nine one
6,891 six thousand eight
hundred ninety one, or six thousand eight hundred and ninety one
6,891.525 six thousand eight hundred
ninety one point (decimal) five two five, or six thousand eight hundred and
ninety one point (decimal) five two five
15,451 fifteen thousand four
hundred fifty one, or fifteen thousand
four hundred and fifty one
183,012 one hundred eighty three
thousand twelve, or one hundred eighty three thousand and twelve
$75 dollar sign seventy five (saying dollar
sign before the number returns the dollar amount without cents)
$75.00 seventy
five dollars
$19.95 nineteen
dollars and ninety five cents
½ one
half
2 1/4 two
and one quarter
5/8 five
slash eight or five over eight
150/85
Dates and Times
Result: What To Say:
Phone Numbers
Result: What To Say:
(954)-555-7700 nine
five four five five five seventy seven zero zero
1-(800)-555-1212 one eight hundred five five five one
two one two
Note: The speech engine places area codes in
parentheses.
Punctuation Marks
Result: What To Say:
. period/dot
. point (say when using numbers)
; semicolon
, comma
- hyphen/dash
? question mark
-- double dash
! exclamation point/exclamation mark
… ellipsis
: colon
’ apostrophe
/ slash
\ back slash
“ quote/open quote
” quote/end
quote/close quote
‘ single quote/open single quote
’ single
quote/end single quote/close single quote
( left paren/open paren
) right
paren/close paren
[ left bracket/open bracket
] right
bracket/close bracket
< left
angle/open angle/less than
> right angle/close angle/greater than
{ open brace/left brace
} close
brace/right brace
@ at sign
# pound sign
- hyphen (use for minus sign)
+ plus sign
$ dollar
sign
= equal
sign
% percent
sign
^ caret (note
if this returns the word “carrot” emphasize the “a” in your pronunciation(kAy-ret)
& ampersand
_ underscore
* asterisk
| vertical bar
Natural Language Modeling Technology
Natural Language Modeling technology is used to initiate sending an email by
giving a command as you would speak it naturally. Our Natural Language Model also uses “best
match” technology to help the computer to recognize names more clearly.
Using the Natural Language Model
To activate the Natural Language Model, you need to
say the command word “Computer.” You
must say “Computer” each time before making a request in natural language. The word “Computer” is an activation command
that tells Speak-to-Mail™ that you are about to send an email using the Natural
Language Model. When the computer
responds “yes”, then you can state your request.
For example –
You: “Computer”
Computer: “Yes”
You: “Send an email to John Doe about next Tuesday’s
meeting.”
In the above example, when you say “Computer”, the
message “Natural Language Model is Active” will appear in the lower center part
of the Speak-to-Mail™ screen .
Next: John Doe’s email address will appear in the
“TO” field, the “Subject” field will read “Next Tuesday’s meeting,” and the
cursor will be placed in the “Body” ready to accept your dictation of the
message.
If you decide to add a recipient after you are done
dictating, or if the Natural Language Model did not recognize all the names you
said, you can add names to the email fields by using the individual commands
for placing email addresses. For
example, say, “Select
(person’s name)” or “Copy to (person’s name)” or “Blind to
(person’s name)” and their email address will appear in the box where you want
it.
Each time you initiate a new email using the Natural
Language Model, the previous email message (if any) is cleared by the new
message. If you have not yet sent your
previous message, it will be erased and you will not be able to recover
it. Alternately, if you do want to clear
your previous message without sending it, you do not need to clear the message
first. Just initiate a new email using
the Natural Language Model, and the previous message will automatically be
cleared and replaced by the new message.
The Natural Language Model recognizes various ways
of saying things. In the above example,
you could have said “Send an email to John Doe regarding (instead of “about”)
next Tuesday’s meeting” or “Email to John Doe to confirm next Tuesday’s
meeting.”
The Natural Language Model will also recognize more
complex statements. For example, you can
say “Send an email to John Doe and Jane Doe with a copy to John Smith about
next Tuesday’s meeting.”
More on the Natural Language Model
The following section contains a more detailed
explanation of the technical aspects of this feature. To get the most out of
using our Natural Language Modeling, you need to understand how it
functions. When you say “computer” it
activates the natural language model and the program waits for a command. There are three parts to the command for
sending an email:
1.
The first part
indicates that you are sending an email and sets the natural language model to
listen for fields in the email and names in the contact list.
2.
The second part is
speaking the fields and names in the contact list. You can select multiple
names and place them in multiple fields.
3.
The third part indicates
the dictation to follow will be the subject of the email.
For example:
You: Computer
Computer: Yes
You: (1)Send an email (2)to John Doe and Alice Johnson with a copy to Jane Smith (3)about Tuesday’s meeting.
Computer: Your email for Jane Smith Alice Johnson John Doe is ready for dictation.
When the natural language model detects the first
part of the command, in this example the words “Send an email,” it will start
to listen for names in the contact list and fields in the email, the second
part of the command. When it recognizes
a name it will place it in the appropriate field that preceded the name (To,
Copy to or Blind Copy to). When the
natural language model detects the third part of the command, in this example
“about” it will place all the text that follows into the subject line. When you are done with your statement, the
program will fill in the fields with the email addresses of the names you
spoke, place the cursor in the body of the email, and prompt you that the email
is ready for dictation.
Examples of the command words that can be spoken to
compose an email in natural language are shown in the following table. The first and third parts of the command can
be used in any combination:
First Part (optional) Second Part Third Part
Send “an email” To
(Contact Name) About
Sending “this
message” (Optional) Regarding
Write “this email” Copy To Concerning
Mail Blind
Copy To Subject
Create Blind To Subject Text
Compose Blind copy Regarding
Make *Regards to
Email *Reference
to
Deliver *Respect to
String modifiers – “With a” “And”
“And a” are used between names in the second part of the command names
only when you are sending an email to more than one person.
*The words “with” or “in” can precede these parts of
the command.
*The Contact Name(s) must be spoken based on the way you have sorted the
contact list. When using our natural
language model, it is best to have your contact list sorted by first name so
you can speak the person’s name naturally.
When the contact list is sorted by last name, saying the names can be
awkward.
Your success in using our natural language model
will depend on many factors that are related to good speech recognition. These include adequate training of the speech
engine, quality of your microphone and sound card, how clearly you speak, and
even the name itself.
Our natural language model, like any other, can only
work with what it receives from the speech recognition engine. If the speech recognition engine has
difficulty recognizing a name, the natural language model will not be able to
use it. If your are consistently having
trouble recognizing a particular name, try training the speech recognition
engine to recognize that name using the “Train word” function or try sending
the email using the standard approach of selecting the names and dictating the
subject and body separately.
If you are using the
Natural Language Model to send an email and the speech engine encounters
similar names and is not sure which contact you want to use, Speak-to-Mail™
will display a dialog box with the possible selections from your contact
listbox numbered in ascending order.
In this example, the statement was “Computer” (pause
to wait for reply) “send an email to John.” The best match technology displayed all
John’s in the contact list.
When you see the dialog box, to select the name you
want, say, “choose or select and the contact’s name.” For example, say, “choose John Doe” or “select John
Doe” to select the first name in the list.
You can also choose the desired contact with a left mouse click.
Known Issues With Natural Language and Speech Recognition
Natural Language Modeling is dependent on the
accuracy of speech recognition itself, and therefore, just like speech
recognition, it is not a perfect technology.
In our programming and testing, we found that the
Speak-to-Mail™ Natural Language Model works very well when the recognized text
is 100% accurate, but errors in recognition can degrade its performance.
Speech recognition engines have a particularly difficult time recognizing names, especially names that are not typical. Long sentences combined with names are particularly prone to a higher level of misrecognition, which cause the Natural Language Model to malfunction.
Pointers for using the
Natural Language Model:
In testing, we have succeeded in sending emails to
multiple recipients and recipient types when all the names are recognized. If you are having difficulty with certain
names, you might want to train them in your speech recognition profile (see
Section – “Training Words”).
It will also help for you to think about what you
want to say before you start speaking.
Do not pause in the middle of the request. This will cause the Natural Language Model to
think you are finished making your request.
Try to use shorter, less complex sentences to make
your requests. The longer and more complex your sentence, the more it is prone
to speech recognition inaccuracy which can result in errors in the Natural
Language Model.
Given this, you can expect the Natural Language
Model in Speak-to-Mail™ to function relatively well with simple statements, but
you may have some difficulty in more complex (unusual names, multiple
recipients and recipient types) statements due to the failures in recognition
from the speech engine. Since the
quality of speech recognition is dependent on many variables, each user will
likely have different results.
We recommend experimenting with the Natural Language
Model and different ways of sending an email in order to find out what works
best for you and your computer.
When you are using the Natural Language Model to
send emails, and the speech engine:
Did not recognize a name correctly and placed the wrong person’s email address in the email.
Solution: Use the mouse and keyboard to highlight and delete the incorrect email address, and either type the correct email address or select it individually from the listbox on the left side of the screen by saying the person’s name or right clicking on their name with the mouse.
Does not recognize a name at all and their email address is not in the email.
Solution: Either select the correct
email address using speech as you would to select names individually, or by
right clicking on the person’s name with your mouse. You can also type the person’s email address
in the desired field.
Does not correctly recognize the subject of your email.
Solution: This occurs when the speech engine miss-recognizes a name
for one that is not in your listbox, and includes the miss-recognized name as part of the subject. Say “clear subject” and re-enter the subject either by dictation or with your keyboard, or highlight and delete the incorrect text with your mouse and keyboard.
SpeakPad™ and
the SpeakPad™ Button
SpeakPad™ was discontinued when CoolSoft, LLC
acquired the assets of Yellow Tiger, LLC, the creator of Speak-to-Mail™ and
SpeakPad™. SpeakPad™ has been replaced
by SpeakToText™ 2.0.
SpeakToText™ 2.0 is a much more robust and feature packed speech recognition program. You can learn more about SpeakToText™ 2.0 and get your copy by going to:
www.coolsoftllc.com/main.asp?product=9
The SpeakPad™ button in Speak-to-Mail™ opens a dialog that will take you to our website where you can get your license and copy of SpeakToText™ 2.0.
TROUBLESHOOTING
SOLUTIONS FOR POOR RECOGNITION
If you are experiencing lower than normal accuracy
in your speech recognition, possible solutions include:
Check your microphone settings. Go to the speech properties in control
panel, and check your
microphone settings.
Make sure your microphone is positioned correctly,
and that the position is
consistent with the position of the
microphone when you trained your recognition profile.
Minimize background noise, and try again.
If recognition was good when you started using
Speak-to-Mail™ and it starts
to deteriorate, closing and
restarting Speak-to-Mail™ will often restore good recognition.
If that doesn’t work, it may be necessary to reboot
your computer to restore the best recognition.
Using the “Train Word” function in Speak-to-Mail™ to
train words that are
consistently being miss-recognized will greatly improve recognition of those words.
TODAY MY VOICE IS NOT AS CLEAR
If you are experiencing poor recognition, Speak-to-Mail™
is a fully Mouse and Keyboard compliant Windows application. Just point to any of your contacts, right
click on that contact’s name, and a pop-up menu will appear, allowing you to
place the contact’s email address in the desired field. Click on any button to perform the
corresponding function. Click on any
field and you can type in email addresses or text. If you are typing more than one email
address, remember to separate them by a semi-colon “;” and a space.
Another solution for poor recognition when your voice is not clear is to spend a few minutes creating and training another speech recognition profile. For information on how to create and train a new speech recognition profile, See Section – “Recognition Profiles.
SUPPORT RESOURCES
Speak-to-Mail™ is designed to be simple and easy to
use, and in most cases, you should not encounter any issues that are not
covered in this User Manual or in the support section of our website. If you do encounter an issue you need help
with, support is available on our website.
To get support, you need to go to the support section of our website at www.coolsoftllc.com and enter a support
request. Support is available only on
the full version of Speak-to-Mail™. To
access the “Request for Support” page, you need the email address you used when
you purchased Speak-to-Mail and your order number, and you need to fill out the
request for support form. If you
purchased Speak-to-Mail from a third party, send an email to support@coolsoftllc.com and we will
reply with an email to tell you how to access the request for support
page. Support is not available on the
trial version.
KNOWN ISSUES
Accuracy
Even with state-of-the-art equipment like a very
high quality microphone and sound card, accuracy with today’s speech
recognition engines usually do not exceed 98%, and such high levels are
available only in very high priced specialized software.
With speech recognition engines like the Microsoft SAPI 5.1 speech recognition engine (which we find to be an excellent engine) and other comparable products, accuracy of 90% to 95% is reasonable. Over time, we anticipate the accuracy of speech recognition technology will improve, but today it is not reasonable to expect 100% accuracy. However, most users, after becoming accustomed to using speech recognition, will find that correcting a few errors is still more efficient than typing text by hand.
Notebook computers
Some notebook computers have low quality sound cards
and the accuracy of the speech recognition on these computers can be
disappointing – especially when you are getting good recognition on your
desktop computer.
Issues Unique to Speech Recognition
Learning to use speech recognition is much like
learning any new skill. It can be
compared to learning to use a computer, learning to type, or learning to use
any software application; training yourself to use speech recognition requires
time and commitment on your part.
Pointers: Speech recognition, unlike using a keyboard, requires that there be little or no background noise, and that other people not be talking while you are dictating. It also helps to keep volume on your TV or stereo turned down so your microphone does not pick up these sounds.
System Resources
Speech recognition uses a lot of system
resources. Having a
faster processor, more memory and ample free space on your hard disk certainly
help. Most of the time, you
should not encounter problems with system resources, but there are times when
excessive use of system resources will cause the speech recognition engine and
Speak-to-Mail™ not to function properly.
If you notice degraded performance in recognition,
here are some possible solutions:
Close programs that you are not using in order to
free up system resources.
Disable your screen saver. Screen savers use a lot of system resources
and can interfere with the performance of other programs.
Close and restart your email program and
Speak-to-Mail™.
Known Issues With Speak-to-Mail™
Shared Functionality
Speak-to-Mail™, by its nature, is dependent on third
party software in order for it to function; namely the Microsoft SAPI 5.1
speech recognition engine, and your email program. There are certain functionality issues that
result from how these third party programs work. Known issues at the time of release are
discussed in this section.
Additional issues that are identified after release may be covered in our support pages at www.coolsoftllc.com.
1. Command not Recognized with Highlighted Contact’s Name
When you place a contact’s email address in one of
the email fields, or ask for the contact’s phone number, Speak-to-Mail™
performs the function and highlights the contact’s name.
Since there is no reason for a contact’s email address to appear more than once in a given email, if you ask to place a highlighted (last selected) contact’s name in another field or ask for the contact’s phone number, Speak-to-Mail™ will not execute the command.
There are two solutions:
The best solution is to ask for another contact’s
phone number, and that contact’s name will be highlighted. You will now be able to select the contact
that was previously highlighted and place their email address in another field
or hear their phone number if you are using Outlook.
An alternative solution is to place another contact’s name in one of the email fields, and that contact’s name will be highlighted. You will now be able to select the contact that was previously highlighted and place their email address in another field or hear their phone number if you are using Outlook.
2. Dictation is not Being Recognized
a. You are saying a phrase and the speech engine does not recognize or mis-recognizes what you are saying. One solution is to try saying the same thing in a different way. For example, instead of saying “Mondays Meeting” try saying “Our Meeting on Monday” instead.
b. If the cursor disappears from the screen while you are dictating, or if you are dictating and your speech is not being recognized, most likely the speech engine recognized something as a command to move to another field. Possible solutions include:
The best solution is to say “body” to return the
cursor to the body.
The “Ready to accept dictation” message in the black
box at the bottom of the screen should come on.
If that does not work, try moving the cursor to
another field, like the
subject field (say “subject”) and
then back to the body (say “body”). The “Ready to accept dictation” message in
the black box at the bottom of the screen should come on.
If you manually place the cursor in the “Subject” or
“Body” fields with your mouse, dictation may not be recognized. The proper way to move to the subject or body
field is by speech. Say “body” to go to
the “Body” field. Say “subject” to
navigate to the “Subject” field.
If the above solutions do not work, then you may have an issue with system resources. Try closing and restarting your email program and Speak-to-Mail™. If that doesn’t work, try closing the programs with the Task Manager. After this, if you are still having difficulty with recognition, save your work and restart your computer. See Section – “System Resources” above for more details.
3. The Cursor Has Disappeared
If it appears that you have lost the cursor even
though Speak-to-Mail™ is still accepting dictation. You can say the command “free cheese” to
bring back the cursor.
4. Installing and Using Multiple Editions of Speak-to-Mail™
You can install multiple editions of Speak-to-Mail™
on your computer, however, you must purchase a license and obtain a valid
registration code for each edition, and you should not have more than one
edition running at a time.
NOTE: You should only use one edition of Speak-to-Mail™ at-a- time. The use of multiple editions of Speak-to-Mail™ running at the same time is not supported and may produce undesired results.
5. NETSCAPE 7.x EDITION SETTINGS
a. Incompatibilities With Speak-to-Mail™ when Netscape 7.x and Mozilla Are Installed on the Same Computer
If you are using the Netscape 7.x edition of
Speak-to-Mail™, Netscape 7.x and Mozilla should not be installed on the same
computer at the same time. If both of
these programs are installed on the same computer, then the system may cause
the Netscape 7.x Edition of Speak-to-Mail™ to read the contact list from
Mozilla. This is because Netscape 7.x and Mozilla are substantially similar.
b. Setting Netscape 7.x to Send Messages As Text
The default for Netscape 7.x is to compose messages
in HTML; however, Speak-to-Mail™ sends messages as text. If you do not change this setting, then
character returns will be ignored and your message will appear as a single
paragraph without any line breaks.
To set Netscape 7.x to send messages as text, go to
Edit, Mail & Newsgroup Account Settings.
In the center of the bottom of the Mail & Newsgroups Account
Settings dialog box, uncheck the box next to “Compose messages in HTML format” .
c. Using the Signature File Settings With the Netscape 7.x Edition.
Both Netscape 7.x and Speak-to-Mail™ can insert a signature file. Therefore, you should enable the signature file in one or the other. If you use the signature file from both programs, you will end up with two signatures in your email.
d. ISSUES UNIQUE TO THE NETSCAPE 7.X EDITION
If Netscape 7.x is open when you start Speak-to-Mail™ , the address book will be locked, and Speak-to-Mail™ will
be prevented from reading the contact list.
In order for Speak-to-Mail™ to read the Netscape 7.x contact list when
it starts, Netscape must not be running and the Netscape Quick Launch Icon in
the lower right hand corner of your taskbar must be closed. After Speak-to-Mail™ has started, then
Netscape can be opened.
To disable the Netscape Quick Launch, go to Edit,
Preferences and in the Preferences dialog box highlight the Advanced
category. In the lower left hand side of
the dialog box, uncheck the Enable Quick Launch box.
TECHNICAL SUPPORT
We have designed Speak-to-Mail™ to be simple and
easy to use. Answers to most of the
questions can be found in this User Manual.
Additional resources may be available on the support pages of our
website at www.coolsoftllc.com. Typically, these self-help resources have
all the answers you need, and are a faster and more efficient way for you to
solve your problem. You should request
support only after you have tried to resolve an issue with the available
resources, and have not succeeded.
A.
Technical Support Policy
CoolSoft, LLC provides technical support on Speak-to-Mail™
only on a best efforts basis, and only to users who have purchased the full
version of Speak-to-Mail™.
We will not respond to requests for support on the following:
·
How to questions or technical issues that are already covered in the User
Manual or in the support section of our website.
·
Issues related to the Microsoft speech recognition engine, or issues with
Speak-to-Mail™ that are caused by the Microsoft speech recognition engine.
·
Issues related to your email program, or issues with Speak-to-Mail™ that
are caused by your email program.
·
Requests for support on the trial version.
· Requests for support that do not have complete information as indicated under “Information Required for Support” below.
B.
Information Required for Support
We will need the following information when you
submit your request for technical support.
Having this all of this information will enable us to help you more
quickly and efficiently.
·
Edition of Speak-to-Mail™ you are using.
·
Your Current Email Address:
·
Your Phone Number:
·
Computer Brand:
·
Computer Model:
·
Operating System:
·
Processor Type:
·
Processor Speed:
·
System Memory (MB)
·
Hard Drive Size:
·
Hard Drive Free Space:
·
Microphone Brand:
·
Microphone Model:
·
Sound Card Brand:
·
Sound Card Model:
The following are of particular importance:
·
Description of the Problem:
·
Exact Steps We Can Take to Reproduce the Problem (our support staff
cannot solve a problem that they cannot reproduce):
·
List Any Error Messages:
·
Steps You Have Taken to Solve the Problem:
C. Support on Third Party Software used with Speak-to-Mail™
If you need help with the Microsoft speech
recognition engine or issues caused by the Microsoft engine, you can use the
help resources in the Speech Properties dialog box under the “Help” tab, (startà, settingsà, control panelà, speechà, other tabà, help); or you can visit the speech section of the Microsoft website at http://www.microsoft.com/speech; or
you can contact Microsoft technical support directly.
If you need support with your email program or
issues caused by your email program, use the help feature in that program, or
contact the manufacturer’s technical support directly.