The coronary heart of the Google AIY Voice set up is the Voice Hat, an incorporate-on board for the Raspberry Pi Zero. The Voice Hat does not really have any onboard speech processing. That is managed by Google’s cloud (or yet another assistance like Amazon’s Alexa). The Hat mostly supplies a decent speaker amplifier and a secondary board with a stereo microphone.
The authentic Voice Hat, working with a full-dimensions Pi, broke out a number of enter-output pins with house to effortlessly incorporate servos as perfectly as travel some greater hundreds (up to 500mA). This makes it effortless to give your voice-controlled gadget some movement or other physical interface. The newer launch uses the Pi Zero, so it’s a minor additional constrained — but it does arrive with the Pi Zero alternatively than possessing to supply your own.
Like the Voice package, the AIY Eyesight package also has an incorporate-on board for the Raspberry Pi Zero, a cardboard enclosure, and an arcade button. But this board, the VisionBonnet, has some actual ability to do onboard graphic analysis without the need of the cloud. It uses an Intel Movidius MA2450 vision chip together with the Raspberry Pi Digital camera Module.
The MA2450 is designed for minimal-ability environments like cellular telephones and allows the Pi offer with the broad sum of facts produced by the camera’s are living online video stream, permitting this modest gadget to system the enter and realize faces and other objects immediately.
Google’s instance code supplies pre-experienced styles for faces, expressions, and objects like cats and canines. You can even practice your own styles, though not on the gadget by itself. For that, you’ll have to have to dive into a deep understanding atmosphere like Google’s TensorFlow. The system of classifying what an object is, from 1000’s of illustrations or photos, is much too intense for these types of a modest gadget to do in a realistic sum of time. Even so, the raw graphic processing of the gadget is continue to highly effective and truly handy if you want to make a responsive vision-based interface without the need of an pricey computer and graphics card hooked up.
The sort issue of the Pi Zero does not allow for the further breakouts found on the Voice Hat, like the transistors to travel substantial hundreds, but it does split out four of the Pi’s I/O pins, ability, and floor so you can connect further inputs and outputs. You could also sooner or later want to produce a sturdier enclosure, as the bundled cardboard set up can use out soon after a couple reassemblies.
The Matrix Voice is the most capable of the a few boards with an 8-channel microphone array and a chip for audio processing. This is the 2nd board by Matrix Labs, preceded by the additional pricey but additional full-featured Matrix Creator.
The Matrix boards use a Area Programmable Gate Array (FPGA) to system the raw audio enter from the 8-channel microphone array by carrying out duties these types of as sounds cancellation and beamforming. Matrix has preprogrammed the FPGA with numerous of the necessary audio algorithms but you are totally free to tinker with them. Like the AIY Voice package, the speech recognition and normal language processing employed to turn users’ speech into useable commands is managed by cloud expert services like Google or Amazon.
The Matrix Voice supports a couple additional features than the AIY Voice, with the two speaker output and headphone jack, an LED ring, and further I/O pins. If you get the version with the ESP32 chip, you can function the board with or without the need of a Raspberry Pi.
Matrix Labs sees their boards as element of a system of IoT products and apps, and they’ve even offered a repository so you can effortlessly incorporate other people’s apps to your Matrix-enabled Pi.
• • •
Using a voice assistant these types of as Google Assistant or Amazon Alexa with possibly the AIY Voice or the Matrix Voice needs some major set up with those expert services. You are going to have to have to reply concerns about the application you’re producing, as perfectly as produce tokens and qualifications that connect your gadget, application, and the various cloud expert services. This system is documented but not especially clear-cut.
In addition, there is some set up necessary on the Pi by itself to configure the components and install the development environments and illustrations. It’s handy to have some knowledge with Linux and/or the Raspberry Pi atmosphere if the build does not go fully smoothly.
The huge gain of all these boards is the preprocessing they do with raw audio and online video enter. And with the audio boards, numerous of the AI features of the cloud expert services like Google Assistant and Amazon Alexa can be accessed from a simple Raspberry Pi computer. So why not give your subsequent job some smarts?