Smartphones, speakers, cars, robots, computers and many other mobile devices surround us every day. With each new model they are more and more improved. These systems perform complex activities that were once associated only with humanity. They implement functions of speech recognition, learning or reasoning.

Artificial Intelligence (AI) is everywhere, making a huge difference in our daily lives. It imitates the activities of the human brain, applying them in technology.

How does it actually works?

The first step in speech recognition is the conversion of spoken words into an electronic signal, which can be divided into categories and processed into action. Initially, the conversion of sound to this signal is done using a microphone. It is then converted to digital data that the computer can understand using an analog-to-digital converter.

Speech recognition is a technology for identifying human speech that the system will convert to text. It is subject of constant improvements and combinations so as to relieve us of activities by touching the mouse or keyboard. All you have to do is enter a command verbally and we will receive an answer immediately.

Through speech recognition, we can distinguish between two main categories. The first is dependence on the speaker, the second is the style of recognition.

Speaker dependence is divided into the system:

• dependent, when the systems are “trained” to recognize the voice of the person or people who trained them,

• independent, can recognize the speech of any trained user, according to the general vocabulary.

We can hear about a dependent system in voice recognition. It’s a technology that identifies a person based on their voice.

We classify speech recognition style as:

• isolated, this means that the speaker must stop between words or commands. Identification of words 0.96s and less.

• combined, recognition of phrases with a length of 1.92s,

• continuous, the system recognizes the voice we use every day.

Smart devices

Technology companies such as Amazon, Apple and Google have been successfully implementing speech recognition technology on their devices for years. Smart speakers are one of such tools. They understand and follow the instructions given by their owners, thanks to natural language processing techniques. A distinction is made between individual speaker models along with the functionality of personal assistants such as Siri, Alexa and others.

Applications for mobile devices

Sometimes we are so absorbed in work that we have both hands busy or we cannot enter a long phrase in the phone search engine because we drive a car. One of the functions artificial intelligence gives us is speech recognition. An intelligent assistant comes to our aid, available for mobile devices and more. These include Alexa, Cortana, Google Assistant and even Siri. These are just a few of them. Users communicate with the device by voice, also by handwriting. The software analyzes human speech and adjusts it to the appropriate command. The virtual assistant can conduct conversations, search for information on the Internet, answer bizarre questions and much more.

Alexa

Amazon has enabled the integration of Alexa functions in devices of various manufacturers, such as Whirpool, LG and several thousand other brands. Currently supports over 30,000 skills.

Google Assistant

Considered as the smartest assistant. It is available on devices such as Android or iOS phones, smart watches, Pixelbook laptops, Android smart TVs / displays and cars with automatic Android support. An interesting fact is that in the first quarter of 2018 Google reportedly sold 3.2 million of its Home and Home Mini devices, surpassing Echo devices powered by Alexa by 2.5 million.

Cortana

Debut virtual assistant published by Microsoft in October 2017.

Siri

Intelligent personal assistant, which is part of Apple operating systems – iOS, watchOS, tvOS, HomePod and macOS. Its first edition was released in October 2011.

A lot of positives

As you know, we deal with a virtual assistant at every step. With this technology, we can forget about the problem for a moment. The device has a built-in function of reminders and tracking commands. In addition, they will help you plan your day, contact us with your friends, and remind you of important meetings. Ułatwiają w codziennych sytuacjach, na przykład odnajdą nieznaną nam lokalizację. Użytkownicy mogą wpisywać polecenia do Google Asystenta, gdy jest cicho w miejscach, takich jak biblioteki.

Google offers a speech recognition tool based on artificial intelligence Cloud Speech-to-Text, which facilitates programmers to convert sound to text, using algorithms for deep learning of neural networks. The tool works in 120 languages and allows voice control, transcription of sound from call centers, processing of real-time streaming or recording of sound. Cortana on Microsoft computers can manage the user’s email messages. Siri ensures the confidentiality and security of information so that searches and queries are not collected for sale to advertisers. It is also worth mentioning programs for converting speech to text. The applications offer the recording of single words as well as longer statements. Thanks to them we will save your valuable time and increase productivity. They are a great solution for people who type slowly on the keyboard. You could mention a lot. Currently, AI applications are very popular.

Upcoming difficulties

There are many obstacles that prevents the device from recognizing and interpreting speech. They include physiological and environmental factors as well as many others. It is difficult to distinguish individual words if the noise of the speech is superimposed on the background noise. It is worth reminding that in 2017 The word error rate in Microsoft’s voice technology was recorded by the company at 5.1%, while Google reduced its rate to 4.9%. Another problem is the lack of recognition of the Polish language by some applications. Currently, Google Assistant became famous at the beginning of 2019 polish version. In addition, using shortcuts and slang does not always result in computer understanding.

Advantages AI

We deal with artificial intelligence at every step. It helps us in everyday activities and brilliantly performs special tasks. One could say that artificial intelligence serves us with a “hand”. Currently, AI plays a key role as a device for everyday use (telephones, food processors), but not only.

As an example of AI, I will give you the humanoid robot of Sophia. It was endowed with artificial intelligence to learn and adapt to human behavior and to work with people. Sophia’s software was designed in 2015 by SingularityNET. She became the first citizen-robot in Saudi Arabia and has given many interviews all over the world.

Risk

But shouldn’t we be afraid? Can the embedded technology turn against us? Do we have full power over it? These issues raise some questions about data leakage, privacy protection. However, it should be remembered that we give control over aspects of life in exchange for the convenience offered. The technology is constantly being implemented and refined. And well. Who would like a chimpanzee to see a picture of an afro-american after searching for the word “chimp”? Risks such as cyber attacks, less secure private data, bad choices made by machines and technologies or loss of skills could also be mentioned. These are just a few of the embarrassing examples of artificial intelligence.

Is it worth it?

This technology is becoming a fundamental part of our lives. The risk is and always will be. After all, machines are still learning, adapting to our functioning, so they can be more personalised to the user. This is just the beginning of an expansion of these technologies. When using artificial intelligence, it is important to remember that these systems are complex, impressive and accurate, but not perfect.