Knowledge base

Speech recognition and The Grid 2

The Grid 2 can use speech recognition in three ways:

1.       Click cells.
2.       Type into the workspace by dictating text.
3.       Dictation using Windows Speech Recognition.

Recognition will be most accurate when clicking cells. In most cases you will be able to use the system with a desktop microphone without needing to train the system to your voice. Methods (2) and (3) allow you to dictate any text into the text workspace, but you will probably need to use a headset microphone and train the system to your voice.

Activate cells by speaking text

In this mode, the recognition engine is only sensitive to the words in the cells. If you have cells labelled ‘Home’, ‘Up’, ‘Down’ ‘Left’ and ‘Right’, you only need to say the words and the cell will be clicked:

– In User settings->Input Settings->Voice recognition, select the Microsoft recognition engine and check  ‘Activate cells by speaking cell text.
– The Microsoft speech recognition GUI should NOT be running.
– To speed typing you can also speak words in Prediction cells.
– If you don’t want to show the text in the cell, set the cell’s description to the word used to activate the cell and clear the cell’s text property.

Type into the workspace by dictating text

In this mode, you can dictate any text either the text or symbol text workspace.

– In User settings->Input Settings->Voice recognition, select select the Microsoft recognition engine and check ‘Type into the workspace by dictating text’.
– The Microsoft speech recognition GUI should NOT be running.
– To use this mode successfully you will need to use a headset microphone and train the speech recognition engine to your voice.

Type using Microsoft speech recognition UI

In this mode you can dictate text into a small ‘preview’ box and make corrections using voice commands. The text can them be inserted into the text workspace.

– Microsoft speech recognition user interface needs to be running to use this mode.
– You can only use this mode with the Text workspace.
– To use this mode successfully you will need to use a headset microphone and train Windows speech recognition to your voice. See http://windows.microsoft.com/en-us/windows-vista/troubleshoot-speech-recognition-problems for more tips.
– This mode should not be combined with ‘Activate cells by speaking text’.

Notes

– Speech recognition is a statistical process. You should never use speech to control cells that may cause discomfort or damage if incorrectly pressed.

– Background noise in the room may cause the recognition engine to drift and performance to suffer.

– If you are having performance problems, try using Windows sound recorder to record a sample of your speech. If you hear interference in the signal try moving the microphone and cables. If you have a built-in webcam or microphone, make sure you have selected the correct input device.

– Be careful the microphone does not pick up output from the speech engine! Some headset microphones are especially susceptible to this form of pickup if the headphone volume is set too high.

– The speech recognition engine adapts to your voice. If the system is being used by more than one person you should create separate Windows user accounts for each user.