Xiaoyi in Wenjie M7 opens a new entrance to Huawei’s full-scenario smart life experience

Recently, there is a breaking news in the field of AI. A software engineer at Google believes that the Google AI model LaMDA has "self-awareness", and published a 21-page dialogue report to prove that, but I still have a little bit of "AI awakening" Doubtful.

Has AI really developed to this point? After all, even the voice assistant, the most widely used AI technology, is often complained by users as "artificial mental retardation". Even in the driving scene where the voice assistant can play its best role, many old drivers around me choose to let it eat ashes.

Yesterday's Huawei summer release, Huawei released the second model of the AITO brand jointly built with Celis, the M7. I also experienced this model for the first time. In the Hongmeng cockpit in the M7, I didn't expect to take me The biggest surprise is not the interior, seats, and audio, but the voice assistant Xiaoyi.

As a senior Huawei user, I am quite familiar with Xiaoyi's abilities, but this time she still brought me a new experience, and I also began to understand the feeling of the Google software engineer, because this AI is too similar people.

AI awakening? Xiaoyi in AITO M7 is too "human"

The adoption rate of the intelligent voice interaction function in Chinese passenger cars has reached 86% in 2021, but in fact many drivers rarely use this function, because those instructions that even children can easily understand, ” AI is difficult to understand.

In 2016, a video of a Shandong driver "warning" the voice assistant in the car is still a popular video at Station B. The driver's elder brother wanted the voice assistant to make a call, but no matter how many times he said it, the voice with a sweet voice The assistant could not understand.

Similar experiences may be familiar to many drivers. From my own experience, at least these situations make me lose patience with the car voice assistant:

  • After getting in the car and starting, if you want to use the voice assistant to set multiple functions such as air conditioning, seat ventilation, navigation, and audio, you have to wake up repeatedly, and you have to wait for it to wake up after the broadcast each time. The efficiency may be slower than manual.
  • When traveling with family and friends, the voice assistant may be disturbed by the sound of the co-pilot and the back seat. If you accidentally touch it or cannot hear the command, you can also make everyone quiet before you issue a voice command, but it is embarrassing to think about it. Bears are also a safety hazard.
  • After waking up the app on the car screen through the voice assistant, some operations still have to be swiped or tapped manually.

In the experience of Wenjie M7, I also focused on testing the performance of Huawei Xiaoyi in the above scenarios.

Multi-sound zone smart perception: family travel artifact

In family travel, you may have encountered such an embarrassing situation. The car is full of young and old, the driver wants to change the song, the co-pilot wants to adjust the seat, the rear passenger wants to open the window, and there are bear children. Yelling, you have to be distracted while driving to meet the different needs of your family.

This time, the multi-sound zone intelligent perception on the M7 can finally solve this trouble. Xiaoyi can accurately identify commands from various positions in the car, such as the main driver, co-driver, and rear row, and the personnel in the car can independently issue commands to Xiaoyi. .

I simulated the scene of a family trip. Different seats in the car were full of people. At the same time, I gave different instructions to Xiaoyi. Xiaoyi can provide services in time according to our instructions, and can also turn off the sound area where the bear child is located. prevent him from making trouble.

To achieve such an experience, excellent software and hardware collaboration is required. First, through the microphone array distributed around the car, Xiaoyi can identify the sound sources in different positions, and then filter out the noise of non-target sound sources to accurately identify the sound source. .

At present, there are actually many cars that support multi-sound areas, but some of them are due to the lack of good adaptation of AI algorithms and software and hardware, the correct recognition rate is not high, or it needs to be very loud or close to the microphone to succeed, and the experience is uneven.

When my colleagues and I experienced Xiaoyi's multi-tone perception, we could basically give commands smoothly with our usual speaking voice and sitting posture, and the experience was quite smooth.

Continuous dialogue: This is the state of talking to people

As mentioned earlier, multi-sound zone perception needs to eliminate interfering sound sources so that Xiaoyi can recognize the commands that need to be heard. When she does this well, she can naturally achieve another practical function – continuous dialogue.

To put it simply, you can directly issue multiple commands to Xiaoyi at one time, without having to wake her up every time, you can also interrupt her in the middle, and Xiaoyi can automatically block invalid commands by replying a voice message or chatting a few words. , Xiaoyi can support 60 seconds of continuous dialogue.

This is a very useful function for drivers. For example, after I get in the car, I set a navigation purpose and then add a waypoint, then turn on the air conditioner, seat ventilation, and finally open the podcast that I didn’t listen to last time, all in one go.

Moreover, this time Xiaoyi has also added a quick command of the main driver without waking up, so you don't need to call Xiaoyi every time you operate, which is more efficient and will not be embarrassing when there are many people in the car.

In the past, I rarely used voice to control these functions in the car. It is not that the car does not support it, but I have to call xx classmates N times, and I have to wait for each command to be executed, which is far less fast than my manual operation.

Visually speaking: use the sound as a mouse and free your hands

Although there are more and more functions that can be controlled by the in-vehicle voice assistant, after opening some application functions through voice, operations such as pulling down pages, switching tabs, etc. often still require clicking on the screen. If the functions are hidden deeper, enter the multi-level Menu search, there are safety hazards in driving.

On Wenjie M7, this kind of inhuman interactive experience has begun to be valued and improved by manufacturers. Xiaoyi supports "visual and talkable", and the functions seen on the screen may be directly controlled by voice, just like using the voice as a mouse to "slide and click" directly, and you don't have to do it anymore.

This function is also adapted to a large number of third-party applications. I tested the more commonly used applications such as Himalaya and QQ Music, which are very smooth.

Not only that, but users don’t have to memorize very accurate instructions. Xiaoyi can understand expressions such as “the first”. They don’t have to spend more time staring at the screen, and they are less likely to be distracted during driving. , this is good.

Xiaoyi's suggestion: really understand me

Speaking of the central control screen, although it is the first time to experience the Wenjie M7, I am not unfamiliar with the operation. This may be because I have become accustomed to using the Hongmeng system on my mobile phone. The mobile phone-to-car experience has also changed a bit.

"Xiaoyi Suggestion" on the mobile phone will remind me of express delivery information on the way to get off work, recommend restaurants and other information according to my location. In Wenjie M7, "Xiaoyi Suggestions" can also judge the user's status based on driving habits and scenes, and give corresponding suggestions.

For example, when the fuel level/electricity is too low, Xiaoyi Suggestion will proactively prompt the nearby gas station/charging station and ask if you need to navigate. Even if you get out of the car and forget to take your phone, Xiaoyi will take the initiative to remind you.

It can be seen that whether it is passive demand or active demand, Xiaoyi has been able to make a more accurate judgment. And the seamless flow of Xiaoyi from the mobile phone to the car makes me feel that the M7's car-machine interaction experience is very similar to that of a smartphone.

Custom Voice: A Human Easter Egg

In this experience, I also found an Easter egg of Xiaoyi that surprised me, which is the custom voice, which means that I can replace Xiaoyi's voice with the voice of my family, lovers or friends, just need to record a voice , Xiaoyi will soon learn.

According to the Microsoft MOS standard, an internationally recognized voice quality evaluation method, the closer the AI's voice is to the human voice, the more comfortable it is. In addition, Xiaoyi also supports the recognition of Cantonese, Sichuan dialect and other dialects, and I finally don't have to quarrel with her with a mouthful of "Boiled Winter Melon".

In the sci-fi movie "Her", the male protagonist fell in love with an AI voice assistant, largely because of the voice of AI. Of course, how many boys can resist the voice of Scarlett Johansson?

Familiar voices can make Xiaoyi appear as an emotional person, the interaction process is more real and natural, and I am more willing to communicate with her. If you turn Xiaoyi's voice into the person you love, even if he is no longer by your side, wouldn't it be a romantic thing?

Moreover, Xiaoyi also supports custom wake-up words, and the overall feeling is more like communicating with friends around you, which seems to reflect Huawei's future voice interaction, which is going to develop towards anthropomorphism.

Although the current AI on the market has not really reached the level of "consciousness awakening", after this experience, Xiaoyi's impression on me is indeed more and more like a human being, at least I can try to communicate with her as a normal person .

This also reminds me of Jarvis, the AI ​​butler in Iron Man. After more than 10 years, we can finally hope to have a real "Jarvis"?

How far is Jarvis from us?

Last week, the European Union passed a proposal to ban the sale of fuel vehicles from 2035, and the auto industry is ushering in its biggest change since the industrial age. The "new four modernizations" (electrification, networking, intelligence, and sharing) have become a trend in the industry, and the automotive industry will change from a hardware battle to a data and system battle in the future.

Volkswagen Group CEO Herbert Diess's judgment on the future of the automotive industry is becoming a reality:

In the future, the car will become one of the most complex but most valuable Internet devices.

In the future, passengers will use the car more frequently, and the time of use may double.

In the future, the car will no longer be a simple box, but a more comfortable and warm space.

The intelligentization of automobiles is an irreversible trend, among which voice interaction has become the core function of car factories to build intelligent experience. With the development of autonomous driving technology, voice assistant will undoubtedly become the most important carrier of human-vehicle interaction in the future, and it can even be said to be one of the new "three major items" in the new car era.

This requires a smart voice assistant that can communicate with you like a human, which is also the impression Xiaoyi gave me. From the point of view of static experience, I think the intelligent experience of Xiaoyi on Wenjie M7 already belongs to the first echelon of in-vehicle voice assistants.

Recently, Xiaoyi has also obtained the first A-level authoritative certification of automotive voice assistant from CATARC. It has performed well in 18 tests including telephone, navigation, car control, visible to talk, wake-up response, semantic association, and multi-round interaction. Excellent, further confirms my judgment.

In other words, Xiaoyi may be one of the current car voice assistants that can communicate with users in the most human-like way of communication.

AI is advancing at a much faster rate than humans, and it shouldn’t be too surprising that voice assistants like Xiaoyi are getting better at understanding you. But this is still far from "Jarvis", because "Jarvis"'s intelligence is not only reflected in his intelligence, but can be deeply integrated and adapted in almost any scene, becoming a universal butler covering all scenes.

In Huawei's territory, this may be the ultimate form of Xiaoyi.

Huawei's new entry for all-scenario smart life experience

In the wave of new car manufacturing , the human-computer interaction experience of the smart cockpit represented by Wenjie M7 is becoming more and more abundant. And these functions are not only serving the driving itself, but are beginning to expand to digital life scenarios connected to the car, such as controlling smart home devices with the car machine, and even giving you suggestions based on the scene of going to work or leaving get off work.

This kind of development seems to be the case when smartphones emerged, and service scenarios other than calls emerged one after another, and mobile phones gradually became an important carrier of people's digital life. Now Huawei may also be able to use the smart cockpit to improve its own interconnected ecology of all scenarios, so that the "mobile terminal" of the car will become an important entrance in the smart ecology.

It is not difficult to connect multiple scenarios, but it is not easy to make the interaction between ecological terminals intelligent and smooth. The interconnection experience of protocols such as Bluetooth and ZigBee used in the past is equivalent to giving fragile single-plank bridges an “isolated island” that cannot withstand the wind and waves, like a chicken rib.

The task of Xiaoyi is to help the interaction of a large number of terminals in the Huawei ecosystem tend to be automated.

In other words, Xiaoyi will be another important entry point for Huawei's full-scene ecological equipment . At present, Huawei has gradually achieved seamless collaboration across terminals and all scenarios through HarmonyOS. In the future, all devices in Huawei's ecosystem can provide users with more scenario-based intelligent services under the management of Xiaoyi, an intelligent assistant. According to the segment, location and behavioral habits, mobilize the relevant device functions to provide the most appropriate service to the user at the most appropriate time.

When the smarter and more "human-like" Xiaoyi seamlessly connects the three scenarios of home-car-carry-carry, it integrates hardware resources, system capabilities, and service ecology, and makes suggestions based on user intentions and executes them efficiently. Then we really have the Iron Man "Jarvis", and the magic of technology has once again brought us into a new era.

Huawei will eventually develop into a "super terminal company" on this basis, just as our previous views on Huawei's future:

Hyperterminal does not depend on the number of terminal types, but on the control of core components and operating systems, the control and influence of the ecosystem, and the forward-looking exploration of human-computer interaction.

#Welcome to pay attention to the official WeChat account of Aifaner: Aifaner (WeChat: ifanr), more exciting content will be brought to you as soon as possible.

Love Faner | Original link · View comments · Sina Weibo