There are several words which may or may not be converted directly into text depending on when and how you say them. For example, if the word “microphone” is said as part of a sentence, it will be converted to text, but if said as a single word, it will turn off the microphone. Similarly, “voice command” said as part of a sentence will appear as text, but said alone, it will switch speech input to voice command mode. Two other crucial commands you can use within dictation mode are “new paragraph,” which is the equivalent of the Enter key, and “new line,” which is the equivalent of Shift+Enter on your keyboard.

If you want to spell out a word, say “spell it” and then immediately begin spelling the word. The letters will appear as you say them until you pause, at which point the system will revert to normal dictation. There is a second option called spelling mode that differs only in execution. If you say “spelling mode,” you must pause before spelling the word. When you start speaking again, the word is spelled out, and when you stop, the system returns to normal dictation. If you want to select a word that has already been converted to text, you can say “spell that.” “Spell that” selects whatever word is touching the cursor at that moment and lets you replace it by spelling it out. This is handy if you see a word translated incorrectly and you think the problem is that it is not in the dictionary rather than it was simply misunderstood.

You must say all punctuation as you go, so the sentence “He said: ‘With this device, I can rule the world!’” would be spoken “He said colon quote with this device comma I can rule the world exclamation point quote.” This system works well except for the odd sentence like: “That’s my final offer, period!” which comes out “That’s my final offer, . !” In these cases, the best thing to do is use the “spell it” command and spell out the word or correct it later. The system also automatically adds a space after a comma or a period and capitalizes the first word of a new sentence.

Numbers follow some special rules. The numbers zero through twenty will be spelled out. Numbers higher than that will appear as numeric values. If you want a number less than 21 written numerically, say “force num” and then the number. You do not need to spell out very large numbers, but you do need to say “point” or “decimal” for the decimal place. The speech “One million two hundred thirty-five thousand and thirteen point five” will appear as “1,235,013.5.”

Most symbols can simply be spoken, and they will appear correctly. The beginning of a Web address is said “http colon slash slash.” Some symbols have special ways of being said that help you get exactly what you want. The dollar sign symbol ($) is best said “dollarsign” as if it were one word. Saying “dollar…sign,” with an exaggerated pause, will result in words “dollar sign.” Table 2-5 lists some common symbols and how to say them:

You may correct converted text with voice command, as well as using Write Anywhere, the writing pad, or the Input Panel keyboard. Voice command lets you perform the Input Panel correction functions and provides some additional capabilities. Figure 2-21 shows the sentence “The voice command mode allows you to make changes using nothing but your voice” with an obvious error. While in voice command mode, saying “select Lloyd’s” would select the incorrect text. Saying “correct Lloyd’s” both selects the text and opens the alternate word list as shown in Figure 2-22. To replace the text with an alternate word off the list, say “select” and the number on the list of the correct word. To delete or respeak the text, read the name of the desired command aloud. Saying “unselect that” would cancel the selection. Voice command also allows you to select all visible text, select after certain words in the text, select text beginning at one word and ending on another, and insert the cursor before or after a specific word. Once you have made a selection, you can also change capitalization in the sentence with your voice.

Figure 2-21. Saying or tapping “voice command” will switch you to voice command mode.

Figure 2-22. Voice command allows you to select, or select and correct text, in a single step. Input Panel displays the command it heard as it carries out the action.

Voice command does much more than give you verbal control of Input Panel. With voice command, you can open and close files, switch between applications, access the Start menu, navigate a document, and more. There are too many commands to describe them all here. To see all the commands available by voice say “What can I say?” while in voice command mode and a list similar to the one in Figure 2-23 appears. What Can I Say is context-sensitive, so the exact content of the list depends on which application you are in and even what you are doing in that application. This feature is invaluable if you use voice for anything more than dictation.

Figure 2-23. What Can I Say is a context-sensitive list of all the commands available in a given situation.

The speech bar has its own options menu, shown in Figure 2-24, providing access to What Can I Say, speech help, and several speech input settings. If you’re having speech recognition problems, the Microphone Adjustment option is quick and can help, especially if you are in a different room than you normally use for speech input. The Voice Training option is very helpful, but each session takes fifteen to twenty minutes. If you like speech input, it’s worth your time to go through several of these sessions. You’ll get the time back in the form of fewer corrections later on.

Figure 2-24. What Can I Say, microphone settings, voice training, and custom word pronunciation are available on the speech bar.

The speech recognizer understands the pronunciations of words based on spelling and grammar rules. Sometimes what it thinks the word should sound like does not match the real word very well, especially for unusual words you added to the dictionary. You have some control over this. Choosing the Add Pronunciation For A Word option opens the window shown in Figure 2-25, containing words you added. After selecting a word, you will hear the speech recognizer’s top choice for the pronunciation. If it isn’t correct, there is an option to record the correct pronunciation. The Record Pronunciation function is a bit misleading in that it does not record your pronunciation and associate it with the word. Instead, Record Pronunciation listens to what you say and matches your pronunciation to its list of possible pronunciations based on spelling and grammar rules. If one of its alternate pronunciations is close to yours, the association is changed and the word will probably be recognized correctly in the future. If your pronunciation doesn’t match one of the alternates, then nothing happens. In my case, the correct pronunciation of Hayabasa (Hi-ya-ba-sa) doesn’t match any alternate, so if I say it during dictation it always converts as “high above the.” If I mispronounce it to match the way the recognizer thinks it should be said (Hi-yab-a-sa), it converts it to text correctly every time.

If you find it annoying to keep switching between dictation and voice command mode, you can make groups of voice commands available in dictation mode. The benefit is you don’t have to switch modes to select and edit words. The downside is your dictation may be misunderstood as a command. To add commands to dictation mode, select Voice Command Configuration, select Working With Text, as shown in Figure 2-26, and tap Details. Check only the groups of commands you want, and check the Enable During Dictation check box.

Figure 2-25. Adding a pronunciation will sometimes fix recognition errors on words you add to the dictionary.

Figure 2-26. You can enable many of the voice commands so that they are available during dictation.

Here are a few more tips to get the most out of speech input:

• Often a combination of speech, writing, and pen in place of the mouse works best for maximum productivity.

• Different room acoustics, low-level background noise, and even your having a cold can interfere with speech recognition.

• If you hide Input Panel, the speech bar is still visible.

• Double check the choice of homonyms such as “there,” “their,” and “they’re”.

• If you’re trying to control an application with speech and it isn’t working, it probably isn’t in the foreground. Tap the title bar of the application window to ensure it is the foremost application.

• If you regularly use speech input in a variety of noise environments, you can create separate speech profiles for each one using the Speech control panel. This will improve recognition overall, but requires training each profile separately in the appropriate environment.

  READERAID2
  Even for a good typist, retyping a document can be tedious. Reading it and having Input Panel do the typing is much more fun and usually works pretty well. Find a printed document that you would like a copy of on your computer. Open a blank word processor document, place the cursor, and activate speech input in dictation mode. Read all the paragraphs, saying “new paragraph” when needed, without correcting any mistakes. Next switch to voice command mode, and save the document with voice commands, switching to dictation briefly to give it a name. Finally, try correcting the recognition mistakes with a combination of voice commands, pen taps, and writing pad inputs.