According to LoupVentures, the most efficient voice assistant is Google Assistant (88%), followed by Siri (75%), Alexa (73%) and Cortana (63%). Google’s digital assistant, in fact, understands the questions in 100% of cases, unlike the other products that have a lower percentage. Bear in mind that the higher the frequency of use, the better the performance of the voice assistant is, thus having more chance of receiving correct and relevant answers. The omnipresence and efficiency of voice assistants like Google have divided humanity in two groups: there are those who love them and those who fear them. Not only Google, but also Amazon, Apple, Microsoft and Samsung have made huge steps forward with their voice assistants, although the various Italian versions are still a few steps behind the US version. In fact, who among us today does not know or has never used the vocal research integrated in the mobile device? Almost nobody. Let’s see together the occasions when the voice assistant from Google or other manufacturers becomes useful, and how it can simplify our daily actions in a way that until a few years ago was unthinkable.
HOW GOOGLE (AND OTHERS) VOICE ASSISTANT WORKS
Voice assistants are systems based on artificial intelligence (AI) and machine learning, able of interacting with people through voice commands. A voice assistant is able, thanks to the algorithms that compose it, to recognize people’s voices, to give the meaning to words and respond to the input received. The performance of the voice assistants changes and improves over time also thanks to machine learning. Speech recognition technology is working to improve the dialogues between people and the virtual operator, which still appear somewhat schematic because they react by choosing the answer from a set of solutions already preconfigured, and not yet completely based on the recognised habits of the Users. The first voice recognition, which is the basis of the voice assistants, dates back to 1952, the year of birth of Audrey (Automatic Digit Recognizer), a device for recognising single spoken digits. Subsequently, in 1964, IBM presented Shoebox, a calculator that solved small calculations directly by communicating them to the device through a microphone. Only from the 1980s were born the first companies that dealt with voice recognition, while in the 1990s voice recognition arrived in the operator’s call centres, to help telecommunication companies manage calls based on the voice interaction of the User. A historical player like Nuance is surpassed by giants like Amazon, Apple and Google. Windows XP and the first mobile phones of the 2000s implemented this technology allowing the voice dictation of SMS or the recognition of contacts in the address book to call. But the real innovations were brought by machine learning and artificial intelligence, which led to the birth of real “handyman” voice assistants, such as Siri from Apple, Google Assistant, Amazon Alexa, Microsoft Cortana and the lesser known Bixby from Samsung. Yet the evolution of voice recognition in the various devices is only in an initial phase, given that the producers will focus mainly on making the dialogues between the User and the software more natural. To configure a voice assistant such as Amazon Alexa and Google Home, first of all you need a device from where you can download the dedicated app and customize it with all the desired features. It is also possible to connect multiple smart devices to each other for the management of household appliances, security systems and much more. Obviously, you will need to have internet connection and a recent smartphone.
Among the utilities of a voice assistant, there is certainly the possibility of asking questions and getting answers, but it is only the first of its functions. Sending messages, listening to news and music via Amazon Music, Apple Music or Spotify or even videos by Netflix and Youtube, following recipes and discovering curiosities, are some of the activities that can be carried out with a digital assistant. Furthermore, the voice assistants are also very ironic: they tell jokes, moderate group games, insert the products to be purchased in the wish list, update the to do list and entertain the Users with conversations on topics of all kinds. And, of course, as it is an “intelligent” system the voice assistant evolves over time to improve. But do users really love these devices? One out of two considers them comfortable and the numbers will increase: four out of ten consumers will use them instead of mobile apps and online sites. A further advantage deriving from the voice assistants is the role they have in the digitalisation of the population, in particular of the elderly who prefer to dictate rather than typing on the smartphone. If these are the advantages, however, there are also some weakness that characterizes this type of technology and that raises doubts in the most sceptical. Cyber security experts have demonstrated how it is easy to manipulate the home assistants, such as the Google or Amazon assistant. There would be more than one way to spy on private conversations and find out the passwords of other Users, therefore the security aspect is still an area whose risks will have to be analysed, in the protection of personal data and people’s privacy. SRLabs, the hacker organisation in question, showed by a series of videos how some apparently innocuous apps can be used with second and subtle purposes. For example, when you ask for something from your smartphone or another device, it may not turn off the microphone immediately afterwards, but keep it open and continue listening to what is happening in the room. Hackers demonstrated how it is easy to change the actions of the voice assistant, to make it send fake error messages, and then ask users to pronounce their password and pass it on to other online services. It is good to know that an assistant would never ask for a password, but a naive User could fall into the trap. These situations are made possible due to the so-called bugs common to Amazon and Google systems, such as the one that allows hackers to add special characters that voice assistants would never pronounce. In this way, the voice assistant continues to read something unpronounceable that the User cannot hear, with the microphone on, to receive any commands sent unwittingly. Obviously, companies have always made sure that the new code complies with the rules before putting it on the market, but until recently, checks were not carried out following the updates of the developers. Today, finally, manufacturers have adopted new solutions to counter the problem, even at the cost of temporarily excluding some features of their devices, waiting to find a solution that respects and protects the privacy of Users. Anyone can install additional skills and actions, more or less authorised extensions, but it is advisable to check the origin of these actions first.
All the main types of voice assistant, from Google to Amazon
The most relevant fields of use for the voice assistant are the following:
For example, the Google voice assistant, it is activated by saying a special command (Okay, Google or Hey, Google). But not only Google, there are also other famous voice assistants such as Alexa, Siri, Cortana and Bixby. Once activated, each of them is able to recognize the words spoken by the User, since it transforms them into text thanks to the speech-to-text functionality. Thanks to the artificial intelligence in the manufacturer’s servers, the voice assistant is able to understand the semantics of each sentence pronounced and to respond to requests made by Users. If all this seems surprising and beyond your imagination, think that all this serves to lay the foundations for even smarter devices that combine voice assistants with artificial intelligence and robotics. We will probably get to deal with robots capable of reacting to our requests and solving problems, therefore going far beyond the simple smart speaker that we use today. Google Duplex, for example, has already shown the potential of the virtual assistant as a handyman secretary who makes appointments on behalf of the User. The fact that artificial intelligence can make calls and reservations and interact with the interlocutor thanks to its synthesised voice is amazing. Google Duplex is still used only by US users, but Translatotron, the assistant that instantly translates from one language to another, may arrive in Italy. Waiting for Duplex, the Call screening feature will soon arrive in Europe, which allows Google’s AI to answer calls and transcribe everything in text form. Of course, one of the situations in which we mostly use the voice assistant of Google or others is when we are driving. Activating the digital assistant, you do not need to touch the smartphone to request something. You can activate the voice assistant not only from the application on your mobile phone, but also from the Android Auto display now abandoned by Google. It provided directions, read messages and replied, played music and managed streaming with the sole use of the voice instructions it received. All these features are now part of the Google Assistant voice assistant. It, as well as all the other most important voice assistants, allows you to manage the smart devices installed in your home and office (does Smart Home tell you something?), capable of connecting to Wi-Fi and enabling features that until recently were unimaginable. Just think of electrical outlets or remotely controlled smart lamps, Wi-fi infrared remote controls managed remotely with an app.
How are the Google assistant and the User’s daily routines connected? You can make sure that the voice assistant, after our Good morning, begins to remember the day’s appointments (with the help of the calendar), which automatically activates the phone’s ringtone, indicates the weather forecast, starts our favourite song and much more. To stay on the subject of music, smart speakers (Google’s voice assistant is a good example) are able to react to the user’s mood by offering the right song for every situation. This is what devices like Google Home Mini, Google Home and Google Nest Hub do, or Apple’s Apple HomePod and Echo Dot, Echo Plus, Echo Input, Echo Show and Echo Spot by Amazon. If you use voice assistants, you will surely have noticed that sometimes they are activated incorrectly, when we did not ask for it. It is useful and necessary to make sure that the voice assistant is not set to be constantly listening, it is right that the User can customise the privacy of his smart device. Lamborghini has also recently chosen to bring the voice assistant Amazon Alexa on board in its cars, providing it with an infotainment system connected to digital services such as air conditioning, interior lighting and seat heating with voice control, with the aim to improve comfort and road safety. In fact, it will also be possible to control the TV and speakers, the home gate, the thermostats and even the lights. To learn more about these topics, visit the following articles: GOOGLE BERT: THE MACHINE LEARNING ALGORITHM GOOGLE CHROME AND THE BLOCK OF AGGRESSIVE ADS THAT CONSUME TOO MANY RESOURCES