Speech recognition is quickly becoming a feature integrated into products like the Amazon Alexa, Microsoft Cortana, and Apple Siri personal assistants. But such speech recognition is not just limited to high-end commercial products. It can also be integrated into even the most basic hobby projects! How can speech recognition upgrade your next project?
Before we continue, it’s first important to define the difference between Speech Recognition and Voice Recognition. It is amazing how many professional companies and engineers interchange these two terms when in reality they are completely different. Speech recognition turns speech into text. For example, a system with speech recognition will hear someone say, “Hello there,” and then display the text “hello there” on the screen.
Voice recognition is about determining who said a sentence. For example, if I said, “hello there,” to a voice recognition system, it would display “Robin is speaking.” A voice recognition system does not care what was said, but who said it. Most systems that claim to have voice recognition almost universally do not, and instead have speech recognition.
Now that you understand the difference between them, let’s look at speech recognition and how it can help upgrade your project!
One of the more useful applications for speech recognition is hands-free interaction with hardware, including even the most basic IoT projects. Buttons and other forms of I/O can be bulky and difficult to use if there are many options available to a user.
For example, if a smart bulb in a room has an RGB LED and display 20 different colors, then using an app or interface to pick a color can be somewhat chaotic. But, if using speech recognition, the user could simply speak the color they want. This hands-free interaction could be expanded further into home-automation systems. Computer terminals - like Raspberry Pi’s - could receive spoken commands instead of needing a physical interface.
Security projects could be made considerably more secure with the use of spoken language as opposed to a keypad entry system. Keypads can be insecure unless a very long password and intricate is used, but this relies on the memory and trustworthiness of the user. A keyboard interface may also be an acceptable password input method, but this too has its own problems. The keypad can be damaged, or an unauthorized user may be able to bypass the software using keyboard commands to gain entry.
A speech recognition system would instead accept a sentence, numbers, or even a short story as a password, and with the lack of a keyboard interface, a hacker must only interact with the system via speech. The lack of a keypad entry also prevents hackers from trying to determine a password by looking at which keys have been pressed consistently.
Speech recognition could also be useful in projects that may require emergency stop mechanisms. Most machinery requires some form of stop mechanism that cuts the power in the event of an emergency. This system could be taken further to any project that may involve a potentially dangerous scenario, such as high torque motors, high-speed blades, high-voltage generators, and even heaters.
If a project is in an environment where an emergency stop system is too difficult to access quickly, then the user could simply shout out a command that would cut the power or apply a break. Of course, such as system should not be relied upon as mechanical switches are almost fool-proof, but a voice-activated emergency stop could still be very useful.
While somewhat far-fetched, a voice recognition system could be integrated into a debugging station. A designer could probe signals with multi-meters and oscilloscopes, while a computer system provides test signals. The user could request specific signals, such as sine waves and pulses with specified features, and then the computer could apply the signals.
The computer assistant could also be tied into an IDE or debug system where a designer could concentrate on the circuit under test. Upon a voiced stop command, the system would halt the current execution of code. A speech recognition assistant could also be made to build projects, erase devices, walk through programs, and possibly even set debug points live.
When it was first released, speech recognition had little success with designers due to its inaccuracies and inability to seamlessly integrate it into projects. Now, with the IoT taking shape and microcontrollers becoming incredibly powerful, integrating speech recognition has never been easier or more accessible. The examples discussed here are only a handful of instances of speech recognition in common projects. Why not consider how speech recognition could improve your own projects?