Speaker type audio assistant who became the biggest hit item in Amazon hardware history "Amazon Echo" The birth of secret story

Speaker type voice assistant terminal that responds with fluent words enough to respond as soon as you talk to it, as if you realized that you were talking to human beings as if you were talking to themAmazon Echo"Is a big hit item in English-speaking countries. Two in three customers in Amazon customer reviews ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ☆ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ☆ I am clarifying.

The inside story of how Amazon created Echo - Business Insider
http://www.businessinsider.com/the-inside-story-of-how-amazon-created-echo-2016-4

Mr. Dave Limp of the Apmazon executive, in the interview of Business Insider, in 2011Amazon EchoI saw that when I saw the prototype of "I think that development of this product will be hard". The user experience that Echo's concept draws was as if it was magical at that time and he felt the need to develop a lot of things until it was realized.

The concept that Amazon has drawn to Echo is a device that responds in response to a voice of a person by voice, such as reading news, putting music, operating home appliances, ordering items at Amazon, so to speak Speech control It was function.

Terminals such as Echo have not existed in the world yet, it is completely unknown whether to succeed by commercializing it. The Amazon research laboratory controlled by CEO Jeff Bezos' perfectionism continued discussion for Echo's commercialization for several years, and it was said that it was stuck at the beginning of the development. furtherFire Phone FailureWith the question mark on the ability of Amazon's hardware development, Echo seemed to have finally made it into commercialization.


Bezos CEO backed up the development team as a strong supporter of Echo from the very beginning when the Echo project was launched, but at the same time, he demanded high quality up to abnormal. Echo's Personal Assistant "Alexa" initially heard the user's voice, the average waiting time it took to get accurate answers from the cloud and respond to users was 2.5 seconds to 3 seconds. So, the development team set the target value of response time to 2 seconds, but Bezos said "We will not get anything without a lot of pain, let's give you pain as a former investment in the development team. "The response time is" 1 second "" It is said that it imposed a high-speed response as a proposition.

Even companies that have worked on voice control technology for decades until then required a waiting time of 3 seconds, and all the Echo development teams, which just finished it, finally lost their words. However, Bezos seems to have driven the team strongly that he should aim for the goal that seemed impossible.

BySteve Jurvetson

After thousands of data analysis tests including not only Amazon laboratories but also speech technology researchers, we succeeded in reducing it to 1.5 seconds or less, far exceeding the response speed of competitors. Further improvements continued until CEO of Bezos got OK, and he said that he was even forgetting that the target was 1 second when he completed the task. "We impose a bold goal value with a clear voice, which is Bezos CEO," recalls Mr. Limp.

The difficulty of developing Echo is not limited to the response speed requirement. The most difficult thing was that there were no past products to be modeled. Certainly there were Apple's Siri, Google's voice search technology, Microsoft's Kinect, etc. as voice assistants, but all of them had screens for users to interact, whereas Echo was completely conversational It is said that the concept was fundamentally different from the point that communication must be established with establishment of the technology.

In order to be able to use only voice, it was as delicate and advanced speech recognition / vocalization technology that as the user feels as if he was talking to a real person, it was required. Therefore, in the development of Echo, a method of evaluating satisfaction by performing subjects' conversation between Echo and a real person as a blind test was taken, and information relating to satisfaction was collected by changing response speed and syntax That's right. One of Echo's developers says that the research goal at the time was experiments of psychology rather than technology.

What I remained after enhancing Alexa's technology was the question of how to create "hooks" that consumers want Echo. As a result of in-house review, the idea that gained approval of 40% was "music related content". Therefore, it seems that Echo has decided to install a large speaker to produce high sound quality.


However, the music content was a double-edged sword accompanied by the danger that Echo could misunderstand that it was merely a "terminal for music players". Especially Bezos pointed out the danger and he was most vigilant, and while showing the understanding of the importance of the music contents, why do you ask the developer about the reason to spend a lot of time to improve the music function I heard he did.

Bezos CEO, who used the prototype model just before the product was completed, produced a misunderstanding among the developers that "Bezos lacks passion for music", the sound of favorite film "Battlestar Galactica" during the test He said that he ordered a truck. As a result of seeking high sound quality, according to a survey of the research company 1100 data, Echo seems to have acquired 25% share in the current American wireless speaker market.

At the beginning of the launch, Echo achieved the initial sales target number in minutes after the initial release, owing to a misidentification of the market reaction. We achieved great success by achieving the base of the first iPhone exceeding 1 million by 70 days in less than 2 weeks. Echo, which continues to constantly enter the top 5 popular products of Black Friday and Christmas Eve afterwards, is now pointed out that it is the biggest hit work holding down Kindle in the history of Amazon hardware.

Echo has a compact "Bluetooth speaker connection"Echo Dot"Built-in battery with mobility"Amazon TapAlso appeared.


Then, Echo who succeeded as developers' enigmatic street music player, after that, various contents such as checking the balance of the bank account, ordering pizza, arranging a taxi with Uber, etc. are added and control of home appliances It fulfills the role of a hub to do, and it is evolving further and expanding the range of utilization.

Echo, which broadcasts the commercial at the NFL Super Bowl in America's biggest sports event, is expected to become Amazon's next big billion bigger business in the scale of one billion dollars (about 110 billion yen) by some experts. Mr. Limp talks about Echo's way, "What Amazon is trying to do is to make a computer on the cloud that can operate perfectly in voice."