Who is mr. Marvin?

In June of this year, we launched a smart speaker with voice assistant named Marvin on the market. It can work up to 2 hours on a single charge. The speaker has six microphones for processing voice commands. The software is fully developed inside MTS without the use of third-party commercial solutions. In this post, we will analyze the main functions and technical characteristics of the new item.







What can Marvin do?



By voice request, virtual assistant Marvin can answer various questions, allows you to listen to more than 60 radio stations, find out the weather forecast, set reminders and alarms, turn on voice games, and learn about the latest news. For the column, you can choose one of three content modes: 18+ - completely free from censorship in relation to songs, news (the content is available in its original copyright), "Without mat" - a mode for adults, but with a filter that blocks specific vocabulary, "Child ยป- restrictions are imposed on the issuance of content for adults (songs without mat, no news, books - fairy tales).



Also, the column is equipped with a built-in Smart Home system module from MTS and supports ZigBee and Bluetooth Low Energy devices. No reseller partner is required to connect the system, no third-party hardware solutions are required. This is the first portable speaker - a smart home control center, which, in turn, is controlled by voice commands, and through the MTS Smart Home application.



What's with the sound?



The speaker is equipped with a digital signal processor for dynamic hardware equalization of the content played by the device. This ensures a harmonious, beautiful and high-quality speaker sound. The DSP coprocessor allows you to control the distribution of the spectral power of the signal, makes the sound dynamically balanced, and controls the distribution of bass. It artificially improves the sound quality by constantly monitoring the signal spectrogram and generates a stereo signal while maintaining the sound effects inherent in this technology. This solution allows you to preserve the sound quality of music and other content without losing the quality of speech recognition.



For harmonious reproduction of audio content, the device is equipped with stereo speakers and a passive bass source. The total average sound power is 10 watts.



The speaker can be connected to any mobile device or laptop using Bluetooth and Airplay as an audio output device.





Picture of the speaker package



About charging



The product is technically unique - despite its small size, it has a battery that, in full load mode, provides autonomous operation for up to 2 hours.



BY



The team of the Center for Artificial Intelligence has integrated software into the smart column that allows you to activate Marvin. The software processes a voice request from an array of microphones at a distance of 4-5 meters from a person when the content is on, the device can operate in noisy conditions. The software was developed inside MTS specifically for a smart device without using third-party licensing methods. The development team managed to create a completely end-to-end acoustic signal processing solution without transmitting data to a third party, which allows you to simultaneously control the speaker with voice commands and use it in Bluetooth headset mode or play music via Airplay.



The smart speaker software has a cloud upgrade function, the user does not need to take any action to download the new firmware version. If a new firmware version is available, it is updated independently when the speaker is connected to the Internet.



Filling



The device consists of three electronic boards and has a high performance, the general specification is presented in the table:







Safety



For owners of any smart device, the security and confidentiality of the use of personal data is important. The smart column is equipped with means of encryption of all data transmitted locally and globally, the data inside the device is also protected.



In addition, users have the option to mute microphones with a button on the top panel. The device does not transmit data in real time, it has a built-in autonomous technology for waiting for an activation request, means for its refinement, and only after starting these stages it begins to form a user's voice request.



Usability



Launching the smart speaker does not require complex manipulations: after installation and registration in the MTS Marvin application, the user can activate the device. During activation, Marvin will dictate a secret code - an additional means of protection during authorization.



The smart speaker can be connected to several Wi-Fi networks at once, while networks of the 2.4 GHz and 5 GHz bands are supported. When connected to multiple networks, software roaming is activated, thanks to which the speaker can be moved between the coverage areas of these networks - the device will switch between them automatically. This functionality makes the use of the device convenient when moving inside a large room, and also provides more stable operation when the Wi-Fi channel is loaded. At the same time, the activation of the column is reliably protected, it does not transmit any data of the acoustic spectrum in the style of a dial-up-modem.



To be continued



The MTS smart speaker is the beginning of a new lineup of the company's smart devices. The "stuffing" and software development of the speaker will become the basis for the next models of an autonomous multimedia gadget with built-in mobile communication and a voice assistant named Marvin. So follow the news and our blog.



Roman Zhukov, Head of AI Devices Development Department, MTS Center for Artificial Intelligence.



All Articles