Let's assume that you need to find some information. How exactly will you do this? Type on the keyboard? Use handwriting input? Or maybe you will use voice search?
By the way, which of the above methods is the most convenient and natural? Perhaps the option with a voice is going to win this race. After all, conversations have been the main way to find out about the desired information for many millennia.
Indeed, recently the voice search is gaining popularity, and the number of voice assistants is growing! We do not know how much this phenomenon will affect our society in the future (for example, the number of children named Alexa), but it is already clear that voice search cannot be ignored in the context of successful promotion of your business. And so that everything works as efficiently as possible, it would be nice to understand the mechanisms by which these digital gears turn. Today we will help you with this.
Simply say aloud “ok MassMedia Group, how does the voice search work?”
It all starts with phonemes.
Absolutely any tool that uses voice recognition, operates with the minimal sense distinctive units of the language - phonemes. To keep it simple phonemes are basic sounds that distinguish words from each other.
As you might guess, several phonemes, one after another, ultimately form a word. So, any voice search service in order to distinguish what you said must first learn to distinguish phonemes himself and absolutely all the sequences in which they can be. And if the first is technically not so difficult, then the second is a real hell for any developer who has tried to do it manually.
Neural networks come into play
And yet, for a long time, it was manual programming that remained the only way to effectively teach a computer to recognize human speech. The engineer had to manually read the sounds, and then translate them into a computer-comprehensible look. Such systems have been around for quite some time, for example, the first system capable of recognizing the numbers spoken by man was built back in the 1950s.
Nevertheless, this method was too inconvenient, and its possibilities were limited, since the computer could only match the existing data. And then neural networks entered the stage. With the help of special algorithms, such systems could self-educate, allowing them to process a massive amount of information much faster and more efficiently, while learning from their own mistakes.
However, one of the most difficult aspects in the recognition of human speech is not the recognition of sounds and their formation into words itself, but an understanding of the context in which these words are used. What is a trivial task for our brain is an incredibly difficult task for a computer.
Although neural networks have made great strides in this industry (Apple showed impressive results in one of its presentations), we are sad to report that existing solutions do not yet understand such things as humor, irony and sarcasm.
Nevertheless, they are already able to learn to recognize the context of simple queries and to distinguish a question from a statement by intonation. However, how to properly use these tools to get the best result?
The game is the same, the rules are different
If you are a business owner, we have two news for you. The good thing is that the general principles of SEO optimization do not change for voice search. The bad one is that the voice search creates many nuances that need to be taken into account, adding even more headache. However, if you are a user, you don’t have to worry at all - everything written above is good for you.
So, requests made with the help of voice assistants have several key features that distinguish them from text search:
1. People use longer phrases for voice commands
It is known that short phrases made of 2-3 words work better with text input. In this case, the person does not use the complete sentence. Just a few words that show the purpose of the search. Due to the fact that the pronunciation of the request takes less time, people, as a rule, are not too lazy to clarify and pronounce longer sentences.
2. Using natural language, a person uses more questions
In voice search, conversational speech is used, and therefore a greater number of question words. Thus, search queries become more accurate.
3. Voice search shows intent
As a rule, uttering a request, a person immediately voices his intentions on the subject of interest. For example, instead of “IT company Ukraine”, the user is more likely to ask “Which company is the best to order custom software?” (MassMedia Group, of course).
4. Voice search is of great importance for local entrepreneurs
Above, we already talked about context. And if the recognition of tone and emotional coloring of the vocabulary still rather vague, modern services manage quite well with understanding of such nuances as “Where does the user need it?” And “When does he need it?”. It is obvious that a person looking for a “Good restaurant nearby” is only interested in places within the city where he is currently located.
Moreover, experts predict, that we are currently on a threshold of voice search revolution and by 2020 at least 30% of all searches will be done without the use of a screen.
And with whom of them should I talk? Despite the fact that this technology began to work relatively recently, today there are a large number of voice assistants. Just look at this list.
And there are even more. Nevertheless, we decided to focus on the 3 most popular ones, namely Google Assistant and Apple’s Siri.
Perhaps you will be shocked, but this is a product made by Google and its main task is to serve as an assistant. Wow! The main advantage of Google Assistant is a convenient integration with the apps ecosystem of the company that created it. And given the popularity of the Android operating system, the search engine from Google and the Google Chrome browser, Assistant becomes a great solution, as it is always at hand.
Google Assistant functions include the writing and management of notes, auto-check of the updates, secure storage of passwords and other information. In addition, the main feature of Google Assistant is a convenient search on the Internet, as the service is able to capture the essence even in a chaotic set of words. Google Assistant showed impressive results at the Google I/O conference, where he independently arranged to reserve a table in a restaurant.
In addition, Google Assistant is integrated into the Google Home smart speaker, as well as into many other third-party devices. The number of languages that Google Assistant is able to understand is also quite impressive.
Siri voice assistant is created by Apple and for Apple devices. It can be considered as its main drawback, however, within the ecosystem of this manufacturer’s devices, it can be used to perform many actions: from a task to turn on the power saving mode, to a request to tell a joke.
The list of Siri functions includes setting up a demonstration of notifications from different platforms, filtering the output of files by date, recognize songs, send messages, make FaceTime calls, and answer common basic questions. However, in more serious cases, this service deals with the duties of an assistant worse than its competitors.
Microsoft voice assistant, focused primarily on Windows operating system. However, it works on almost any popular device. The strengths of the helper lie in the fact that Cortana seeks to understand the living spoken language and allows to use it to control smart gadgets. For example, in response to the phrase that the room is dark, she will turn on the light, and she’s also able to understand that the phrases “I need a taxi” and “call me a taxi” require the same action.
Another strong point is the capabilities of a personal assistant. For example, Cortana can independently look through all your mail, and then read most important letters. At the same time, the number of languages Cortana supports is rather small, at the moment there are only 8 languages available.
Finally, we decided to check out 3 assistants personally. Watch the results in the video above, if you still haven’t and subscribe to our YouTube channel.