Such obedient cars are rare! No wonder 93.2% car owners love to use it.

Up to now, the core competitiveness of automobile has gradually changed from mechanical hard core to human-computer interaction, and Xpeng Motors obviously knows this.

Not long ago, Xpeng Motors released a brand-new speech recognition system, which was first installed on the flagship model P7. The professor was lucky enough to be invited to try it at Xpeng Motors Tianhe Headquarters.

What is the best interaction scheme for smart cars?

Let me talk about my understanding of human-vehicle interaction, which can be roughly divided into three categories:

First, based on the interaction of physical keys, it only supports a small number of command sets, such as manual adjustment of windows, air conditioning and other most traditional basic operations, which usually appear on beggar models of classic cars or cheap scooters;

Secondly, based on the interaction between the touch screen and physical buttons, some applications can be operated through the large screen of the vehicle, but high-frequency systems such as air conditioners keep physical buttons, which is convenient for drivers to do blind exercises while driving, and it is also the usual human-computer interaction scheme for most vehicles at present;

Thirdly, based on the interactive mode of rich application ecology, with the built-in application and menu classification of the central control giant screen becoming more and more rich and complex, speech recognition has become a relatively convenient interactive mode for people and vehicles in contemporary smart cars, allowing drivers to focus on driving while safely operating in-vehicle functions.

The third point is the science and technology sector that Xpeng Motors spared no effort to cultivate, and its phased achievements have been reflected in the flagship model Tucki P7. The following is "the world's first all-voice in-vehicle system".

The world's first system-wide functional coverage

The highlight of Tucki's brand-new speech recognition system lies in its wide interactive range. In short, all functional interfaces in Tucki P7 intelligent cockpit can be operated by voice commands as long as you can see them with naked eyes.

At this moment, the text becomes a mouse cursor, which completely replaces the tedious process of reaching out and touching the screen, and truly makes sense.

This also means that the rich and varied menu bar in the train system is no longer a stumbling block for users to give instructions, so they can speak directly. Even if you experience Tucki P7 for the first time, you can enjoy all the functions of the intelligent cockpit (attention, everything) and avoid the embarrassing situation of trial and error.

To realize this seemingly impossible hard-core function, it is inseparable from the trillion-level voice instruction library. Therefore, Xpeng Motors has chosen to cooperate deeply with mature third-party organizations, thus making Tucki P7 so considerate in a short development cycle.

It should be pointed out that if you want to issue instructions through voice commands in third-party applications such as WeChat and iQiyi, you have to wait until the software provider adapts to Xpeng Motors. As for when it can be realized, that is another story.

Speaking of experience, the actual experience shows that even with the intervention of many external factors, such as opening windows, increasing the air volume of air conditioning, playing songs and so on. The voice command recognition rate of this system is still quite high, and the cost of human-vehicle interaction is minimized.

Unfortunately, this voice interaction system only supports Mandarin at present, but not local dialects. Engineers said that in the future, relevant teams will continue to develop and let users taste the fun of communicating with Xiao P in dialect through OTA remote upgrade.

The world's first full-scene continuous dialogue

As we all know, most mainstream models with voice recognition function in the market need to activate the voice control function by calling wake-up words such as "Hello, * *". If you want to activate it again in the future, you have to say the wake-up words again. And some high-end models can appropriately extend the time of waiting for instructions.

On the other hand, Tucki P7 omitted these tedious instructions of "wasting saliva". It just needs to say "hello, little P", and then it can continue to give instructions. Even if there are "digressions" unrelated to instructions, they will be filtered by the system intelligence, and there is no need to wake up Xiao P repeatedly. Constant listening won my heart.

(Xpeng Motors AI product director? Wang Tianyi? Complete patient presentation)

In addition, the voice interaction system also supports semantic interruption, that is, callers can interrupt unfinished instructions in the middle and directly say new commands for Xiao P to execute.

For example, when I ask Xiao P what the weather is like in Guangzhou today, when Xiao P replies "It's cloudy in Guangzhou today ...", I can directly interrupt: "What about tomorrow?"

You don't have to wait for Xiao P to finish broadcasting before asking questions. You can continue to make up the knife: "What about the day after tomorrow?" "What about the day after tomorrow?" "What about Beijing?" Wait, it feels like we're talking to a real person. We can tell the difference.

More than that, the system can lock the sound of the speaker in the main driver's seat. When the main driver says "Hello Little P" to wake up the system, the next execution instructions will be subject to the main driver's, and the instructions issued by the co-driver will be automatically filtered, emphasizing pertinence.

Even more intimate, Tucki P7 is equipped with a voice call button on the back door panel. When the back passengers press this button, they can send out voice control commands and enjoy the same voice interaction experience as the front passengers, which can be described as rain and dew.

In addition, the owner can also define voice commands and corresponding operations at will through the mobile app. For example, if the command is defined as take-off, a series of operations can be realized in the car, such as folding the rearview mirror, flattening the seat, and the maximum air volume of the air conditioner. Custom romantic mode can play love songs, adjust the tone of atmosphere lights and so on.

What's more, Xiao P's image dressing even supports personalized customization, and can be switched to Kawaii images such as goddesses, pets and children. After all, concrete things will make people more eager to understand it. Generally speaking, there is nothing that Xiao P can't do, only what you can't think of.

Although the speech recognition system has been widely used in different levels of vehicles, I believe that there are not many car owners who really use the car voice control function at high frequency. It's not that users don't need it. It's just that the experience of most car voice control functions is not ideal at present, but I think Tucki P7 will be an exception.

Data don't lie. The number of effective voice commands of Xpeng Motors has increased from 65,438+02 times a day in 2065,438+09 to 25 times this year, and the average daily usage rate of P7 voice in Tucki is as high as 93.2%.

The enthusiasm of Tucki users to use speech recognition function is closely related to the exploration spirit of Xpeng Motors related teams in the field of user experience regardless of trial and error costs.

Therefore, this persistent spirit of tackling key technical problems is the key for the new forces to express themselves in the domestic new energy market that keeps pace with each other.

This article comes from car home, the author of the car manufacturer, and does not represent car home's position.