Send commands to an instrument by voice.
- Connect. Press Connect and choose your instrument in the dialog of the browser.
- Register. Enter a keyword and the command to send when the keyword is heard. Up to three pairs. They are remembered in this browser.
- Listen. Check the language (it starts with the language of your browser) and press Start speech recognition. Allow the microphone when the browser asks.
- Speak. Say a keyword. Its command is sent to the instrument, and the log shows what was heard and what was sent.
Try it safely
Turn on Test mode to see which commands would be sent without sending them. Check that Listening is shown: that is when the microphone is open.
Notes
- Speech recognition needs a browser that supports it, such as Chrome or Edge. The audio is sent to the speech service of the browser (Chrome: Google, Edge: Microsoft). Do not use it for confidential work.
- By default only a finished phrase counts. React before the phrase ends is faster, but may react to a half-heard phrase.
- In languages written with spaces a keyword must be heard as a whole word: “on” is not found in “one”.
- If several keywords are heard in one phrase, the one heard last is used.
- The commands are only written. Answers of the instrument are not read, so use commands that do not produce an answer.
- Be careful with commands that switch an output on. Background talk can be mistaken for a keyword.