Talking to Google Voice Search Through Arduino

Posted by


This is a Sponsored post written by me on behalf of Google. All opinions are 100% mine.

Imagine asking your computer a question verbally and having it give you an answer, or at least help you find the answers you’re looking for. This one of the founding principals behind Google’s new “Voice Search” project available to users of the Google Chrome browser. While we may still be some time away from recreating Tony Stark’s JARVIS system, the potential of building something like it may be at hand.

With the Android operating system, Google introduced a form of hands-free search by integrating speech recognition into the keyboard and search bar. This allowed users to enter in large amounts of data or search for something very specific without having to deal with the hassle associated with the smaller keyboard available on a mobile device.

This isn’t the first time speech recognition software has allowed users to interact with their software. Nuance (the makers Dragon speech recognition software) has dominated that realm for some time, with Microsoft making great strides itself with the technology in Windows 7. The one trend they have in common is the need to be trained in order to properly interpret words in to text. Late in 2010, Google added speech training to their voice search feature on Android in order to improve recognition in cases where accents and personal inflections require tweaking on the part of the program. For the most part, Google’s preset speech patterns work surprisingly well, and they’re getting more accurate as time goes on.

More recently, Google has moved their popular speech recognition platform to the desktop by adding the familiar microphone logo to the Google search bar. Currently, this feature is only supported for users of the latest version of the Google Chrome browser. It works by using your default microphone to allow you to say what you’d like to search for rather than requiring you to type it in. This can come in handy in cases where difficult spellings or lengthy search strings may otherwise hinder your results.

So, does it work? Surprisingly, it works very well – even when the microphone is a significant distance away from the person giving the command. No training of the voice search is necessary and the search is completed just as fast as it would have been if I had typed the string in myself. Unfortunately, Voice Search is currently available for users speaking U.S. English; if that’s not your native tongue, you may be waiting for Voice Search. It may also have difficulty working in cases where the user has a particularly thick accent or a high amount of background noise. The technology isn’t perfect, but it’s a good start.

So, how can we expand on this technology and make it work for us in more interesting applications? The folks over at BreakfastNewYork have adapted this open source technology in a way that allows them to create some rather interesting and unique object-based hacks. The Verbalizer is an open source board designed to work specifically with Google’s Voice Search for desktop. In a sense, they went out of their way to make voice search more fun – and it looks like they may be on to something.

This board allows you to try your hand at making something physical to interact with Voice Search. It connects to the system through Bluetooth. When triggered, it opens a new tab in the Google Chrome browser and activates Voice Search. An on-board mic picks up your speech and sends it over to Google, completing the search. The board is Arduino (an open-source electronics prototyping platform) compatible, which means it is fairly easy to find guides on how to modify and expand with a design of your own. If you want The Verbalizer, you can actually sign up through their web site and download the necessary software, documents, and schematics to get started on your project.

Cloud-based speech recognition software is certainly showing signs of its massive potential. Where the technology goes in the short and long term will no-doubt be determined by how users choose to support it. Natural interaction with our computers is another step in the evolution of technology that could lead towards that perfect science fiction scenario where talking to your computer will feel as natural as having a conversation.