Alexa retain users in view and leads group language

Alexa retain users in view and leads group language

Amazon’s announcements for the latest generation of his smart speakers and displays sounded first little spectacular: The new ECHO offers like the ECHO plus an integrated smart home hub with support for ZigBee and Bluetooth Low Energy. He also has – as the new echo dot and the new echo dot with clock – "sphyan" Design, so rather a spherical shape.

The Echo Show 10 (with 10-inch monitor) is in turn in the third generation with integrated Smart Home Hub, which supports ZigBee and Bluetooth Low Energy (BLE) and, for example, routines allowed to automatically turn on the light when someone enters the room.

But the latter Smart Display has it in itself: thanks to the engine and tracking function, the user is always located in the field of view of the new 13 megapixel wide-angle camera. The device recognizes the case of an analysis of the audio data from its microphone array and video data from its camera, from which direction the integrated language assistant Alexa is addressed.

The whole room in view

The engine was also allowed to use to turn the camera when you access a Live Feed of Echo Show 10 from another echo show or over the Alexa app to a (according to Amazon). Since the Alexa app can also be zoomed about the Alexa app, the entire room should be overlooked. And also the video telephony function wants to expand Amazon: customers should soon be with the new function "Group calls" create a group of up to eight friends and family members and the conference with a simple command like "Alexa, call my family" start.

Echo Show 10 will be relevant to the list price of 243.69 euros in the colors anthracite and the woman. A release date not yet called Amazon, but interested parties can be notified about the ticket start.

Alexa retain users in view and leads group language

The new ECHO show 10 is equipped with a burdenless motor that turns the 10-inch display (according to Amazon soaceless).

Conversation round

Amazon also aimed at a small video clip a look into the future of Alexa: in a demo video took the command "Alexa, Join Our Conversation" Part of a conversation with two women who discussed what pizza should order them – and who only turned from and to the language assistant, but without using the activation word again. Nevertheless, Alexa recognized when she was addressed and when not.

The Chief Scientist of the Alexa-Ki Department of Rohit Prasad succeeded in C’t, that this works, among other things, because the sentence construction and emphasis can be seen whether a person is directed to the language assistant. In addition, people were invigorious in the direction of the Echo Show, if they had Alexa something to say. Therefore, the system also values the viewing directions of the talkpellers.

Possibly this makes Amazon’s new "AZ1 Neural Edge Processor", the manufacturer uses in the Echo Show 10. In comparison to the currently used chip, it should process twice as quickly as possible, but only to a twentieth of electricity consumption and 15 percent of memory requirements.

Speaking to three persons possible

According to Prasad, Alexa can currently participate in talks with up to three people. This limitation is above all together with the fact that the amustere level is still considerably increasing in the case of coarse groups and it becomes more difficult (even for humans) to follow the conversation when more people are confused.

The AZ1 will therefore be a prerequisite for the conversation mode according to the current state; A firmware update that refuses the mode at the age of echo show models is not to be expected rather. Unclear remained until the editorial deadline, whether the AZ1 also in the "Kids edition" the echo dot is used. She should recognize whether a child talks to her and, as a result, only offer suitable content.

The conversation mode and the Kids Edition of the ECHO DOT will first be offered only on the US market. A comment on how much time the localization will claim for the German market, Amazon did not give.

In the online single interview, the Chief Scientist answered the Alexa-Ki Department Rohit Prasad (below) questions about the coming development of the language assistant.

Data protection offensive

When presenting the ECHO SHOW 10, Amazon pointed out that the processing of the data for the tracking function is located locally on the device and the data protection is therefore ensured. Users can also turn off the movement function at any time by filling the integrated camera cover, adjusting the settings on the device or in the Alexa app or just the command "Alexa, turn off movement" give.

But also to expand the settings for data protection of his language assistant in general. So users should receive the possibility to contradict a storage of voice recordings. Voice commands to Alexa are then lolled directly after processing, the manufacturer promises.

Premature loaves possible

In the first activation of this option, all previously saved recordings should be removed. Amazon according to own information only transcriptions of the previous requests for 30 more days. On request, users could also let go of them prematurely.

So far, Alexa users had to choose the possibility of language records individually or depending on the device. In addition, there was an option to set all the recordings manually or set a loschrythm of 3 or 18 months. These options remain preserved according to Amazon.

With the conversational mode, for critics, the language assistants like Alexa have already been creepy anyway and have the feeling, the language assistant was recorded their talking and monitoring them, nevertheless exceeded a red line.

This article is from C’t 22/2020.

Like this post? Please share to your friends:
Leave a Reply

;-) :| :x :twisted: :smile: :shock: :sad: :roll: :razz: :oops: :o :mrgreen: :lol: :idea: :grin: :evil: :cry: :cool: :arrow: :???: :?: :!: