Computer learned to listen to the sounds of the TV
Researchers from Google Research have created a prototype system that tracks the sounds in the user's room to show him contextual advertising via the Internet.
The inventors suggest that the best use of this technology is to “listen in” the sounds of the TV. It is known that a very large number of users combine surfing the Internet with watching TV. For many it has become customary to leave the TV on as a “background”.
Marketers have long been looking for ways to use this phenomenon for their own purposes. For example, often at the end of a television show, a website address appears on the screen where the viewer can go immediately. The integration of television with the Internet takes on various forms, but Google’s new invention is a qualitatively new level of symbiosis between the Web and TV. Oddly enough, the idea is very simple. Through the built-in microphone to listen to sounds in the room. The analysis of this information allows us to understand which channel and what program is currently on TV. After that, the user is automatically received via the Internet related information: services, products from online stores and various advertisements associated with this TV show.
')
Researchers from Google talked about their work at the Euro ITV European conference on interactive television, and even received an award for the best report (PDF file), which was proudly reported on in the Google corporate blog.
As follows from the description of the invention, the system does not directly transmit audio or speech recognition recorded through a microphone. Instead, the information is transformed into a spectrogram, then analyzed and encoded in the form of special 32-bit "tags", which contain some statistical information about the recorded sound of the TV show. One label corresponds to one fragment, and the duration of the fragment may be different, for example, five seconds. Using these tags you can determine which program is on TV.
When the transfer is identified, in the second stage, you need to determine which contextual information should be transmitted to the user. Researchers offer four options.
1. Personalized information. For example, information about what brand of clothes is not on the main character of the film, and where you can buy it in your city.
2. Direct links to the community. If you want to talk about this TV show, then you will see a list of your friends who are also watching it at the moment.
3. Popularity ratings of TV shows in real time. Sociological services conduct such studies of the popularity of programs, but there a) special equipment is needed; b) research is very expensive; c) results are not available in real time. The system, invented by Google, is devoid of all these shortcomings.
4. TV bookmarks. The user can click on the button to save the given TV show in his video collection.
Experimental testing of the new system was successful. As part of testing, the laptop was placed a few meters from the TV next to loud-speaking people. Even in such conditions, the system successfully recognized the telecast and produced relevant content.
It is not yet known whether it is possible in the future to use a new invention for analyzing human speech, for example, talking to people in a room to show ads based on this information.
The inventors suggest that the best use of this technology is to “listen in” the sounds of the TV. It is known that a very large number of users combine surfing the Internet with watching TV. For many it has become customary to leave the TV on as a “background”.
Marketers have long been looking for ways to use this phenomenon for their own purposes. For example, often at the end of a television show, a website address appears on the screen where the viewer can go immediately. The integration of television with the Internet takes on various forms, but Google’s new invention is a qualitatively new level of symbiosis between the Web and TV. Oddly enough, the idea is very simple. Through the built-in microphone to listen to sounds in the room. The analysis of this information allows us to understand which channel and what program is currently on TV. After that, the user is automatically received via the Internet related information: services, products from online stores and various advertisements associated with this TV show.
')
Researchers from Google talked about their work at the Euro ITV European conference on interactive television, and even received an award for the best report (PDF file), which was proudly reported on in the Google corporate blog.
As follows from the description of the invention, the system does not directly transmit audio or speech recognition recorded through a microphone. Instead, the information is transformed into a spectrogram, then analyzed and encoded in the form of special 32-bit "tags", which contain some statistical information about the recorded sound of the TV show. One label corresponds to one fragment, and the duration of the fragment may be different, for example, five seconds. Using these tags you can determine which program is on TV.
When the transfer is identified, in the second stage, you need to determine which contextual information should be transmitted to the user. Researchers offer four options.
1. Personalized information. For example, information about what brand of clothes is not on the main character of the film, and where you can buy it in your city.
2. Direct links to the community. If you want to talk about this TV show, then you will see a list of your friends who are also watching it at the moment.
3. Popularity ratings of TV shows in real time. Sociological services conduct such studies of the popularity of programs, but there a) special equipment is needed; b) research is very expensive; c) results are not available in real time. The system, invented by Google, is devoid of all these shortcomings.
4. TV bookmarks. The user can click on the button to save the given TV show in his video collection.
Experimental testing of the new system was successful. As part of testing, the laptop was placed a few meters from the TV next to loud-speaking people. Even in such conditions, the system successfully recognized the telecast and produced relevant content.
It is not yet known whether it is possible in the future to use a new invention for analyzing human speech, for example, talking to people in a room to show ads based on this information.
Source: https://habr.com/ru/post/4077/
All Articles