This page has been robot translated, sorry for typos if any. Original content here.

We speak with a computer

Everyone must have observed that in the films the characters are talking to the computer with might and main, moreover, he gives them a meaningful answer. Now programmatic progress has made all this available to every user. If there is a processor from 1 GHz also a microphone, it is allowed to train Windows to perform various voice commands, and in any applications. Suppose you are allowed to make commands that simulate keystrokes, but then use this “feature” in toys: for example, click “1” on the phrase “Choose a knife”, and “2” on the phrase “Choose a gun”.

So, let's begin. First, let's deal with the microphone. I note that for use in toys it is more convenient to use a microphone, which is attached to the head - this bundle works more clearly, and it also does not interfere with various noises. And to control the computer, obviously, a desktop microphone is better, so that it was allowed to give commands without coming up to the computer.

For owners of the popular Creative SBLive! 24bit I note the fact that the microphone input there is more intelligent than on other sound cards. If you use an unsophisticated condenser microphone (having just a couple of wires), it also doesn’t work for you at all, then in its connector you should connect pin L to pin R (if it has a “mono” connector, you must first solder it to “stereo”) . Otherwise, the microphone input operates in low sensitivity mode. It is impossible to use the microphone modified in this way with some other cards - for other cards, feeding for electret microphones can be displayed on the R-contact.

If you eat two sound cards, for example, a regular PCI also integrated on a mother’s reward, then it will become more rational to use an integrated sound card for a microphone. Since on the main card the mixer can be periodically reconfigured for other purposes (to the sample, for recording for broadcasts from the tuner). And an additional card is allowed to be allocated to exclusive ownership of the voice control program.

The microphone must be set up first in Windows. Double-click on the mixer icon in the tray, the mixer appears. There are a couple of mixers for each sound card in Windows - one for playback, the other for recording. We need to configure the one for recording. To do this, go to "Options -> Properties" and select the "Record Settings" menu there.

Allow display of the "Microphone" control. Then in the parameters of the first window of the mixer we put a daw on "Advanced parameters".

An additional “Setup” button will appear under the microphone control, go there.

Usually there is the option of additional amplification of the microphone signal, for example, "+ 20dB Boost" or "Booster" 1. It should be activated if you use a desktop microphone, but not attached to the headphones.

Now let's get down to the main program. So, we will use Game Commander 2. Despite the game name, this is a very developed program also with toys in no way connected. Download, install 2.

We go to the settings (for this we press the button with the prompt "Edit").

In the settings, there are already a lot of presets for different toys, but we do not need them at all, since they use letter-by-word recognition of the English voice. All of them are allowed to be removed. Then we make our way into the microphone settings "Tools -> Microphone level".

Here you will have to choose the location of the Volume control so that “Speech detected” is written at the top of all pronounced phrases. If the microphone is desktop, then most likely you need to set the control to the "High" location. However, then, in actual use, also at an extremely high level of sensitivity, Game Commander can begin to confuse teams, making false positives. In this case, reduce the Volume level for the microphone. Close the microphone settings.

Now we will create phrases processed by the program.

The program owns a pair of action types - Global Commands also for specific programs. Actions (more precisely, imitation of keystrokes) from the Global Commands list are done in any active window, as if you would have pressed a key directly at that moment. In most cases, this is inconvenient as well. And actions for a given program are performed only if this program is also running its window is active.

Let's practice on Notepad for the source. First, run Notepad, then in the settings of Game Commander go to "Add -> Command File ..." and select notepad.exe from the list of processes.

Then in the settings log, open a new Notepad branch, expand it and add Command in the Command Set.

On the General tab that appears on the right, select the Voice Training mode in Voice input.

A prompt appears to dictate the key expression of the command.

It is necessary to press Start, monitoring the countdown, dictate the command three times.

Then again on the General tab, in the Insert keystroke into action list field, enter one or more keys, pressing which will need to mimic.

Done. Today it’s allowed to add a few more commands to the Command Set to react to other words spoken. On the Feedback tab of a separately selected command, it is allowed to enter sound files that will be played back until the command is also executed later. You can, for example, write in them the phrases “the team accepted” and also “completed the command”.

And on the Feedback tab of the entire settings profile it is allowed to specify sound files that will be played when the command is recognized or not recognized.

On the Audio Devices tab, you can choose which sound card to use the microphone from - this allows you to use it in no way on the main card, but, say, on an additional one. The Check audio setup option allows you to restore the mixer settings if they were changed by some other program or fun.

Everything, having configured, save the changes, also close the settings window. Now let's start exploring the first window of Game Commander.

The first button - Listen - globally allows the program to listen to also execute commands. Another button - Audible Feedback - allows the application to respond (with predefined sound files) to your commands. The third button - Global Commands - allows the execution of commands from the global list.

Turn on Listen, start Notepad. We pronounce the previously given command. Please note that Game Commander will only accept (and execute) the command if the Notepad window was active. If blah blah he scribbles Not recognized - this doesn’t always mean that he didn’t recognize the command. This suggests that he did nothing, for example, because there was no window at all for which the spoken command was. But the commands from the Global Commands list will be executed in any active window, if, of course, the Global Commands execution button is pressed.

It is allowed to use global commands for universal computer control, for example: switch tasks, turn off the computer, start a web browser. Game Commander does not know how to run programs on its own, it can only simulate keystrokes. But this problem can be solved with the help of MediaKey. First, in Game Commander, we configure the simulation of pressing, for example, LShift + LCtrl + A, but then we prescribe this combination in MediaKey and assign a file launch to it there.

Commands associated with programs, as was also intended by the creators, can be most used in games3. Many different actions are allowed to be assigned to voice commands: choosing one weapon or another, showing a card, reloading, repairing, and much more (in modern toys, as a position, each key on the keyboard does something).

Now that you understand the principles of using Game Commander, you can find her application.


1. For Creative SBLive! Due to an error in the driver, 24bit (and, obviously, some other sound cards) is allowed to be turned on only in the microphone settings in the playback mixer, but it also affects the operation of the recording mixer.

2. Recall that in free mode it has significant limitations: it is impossible to simulate successive keystrokes of several keys; it is also impossible to "press" keys globally (only in pre-specified programs).

3. We emphasize the fact that software as well as technological progress, long ago, already allows making interactive voice control of a computer everywhere, but directed support from application programs is required. For example, imagine a dialog like this:
- Computer: received new mail.
- Man: read it.
- Computer: (reads).

For a long time, such popular programs as TheBat !, ICQ, Miranda can, in one way or another, announce the receipt of mail by voice as well, if they want, they are able to pronounce it (with the help of third-party plugins). But they can’t act on any additional event (like, for example, a person’s team in the presented dialogue), if they read, they read immediately upon receipt.