Apple HomePod Technology Interpretation: Why Does Apple Value Smart Speakers?

OFweek smart home network news Beijing time at 6 o'clock on the June 6, 2017, the 28th WWDC Apple Global Developers Conference, Apple finally released the HomePod finale, a smart speaker Apple carefully built. However, it is estimated that this has disappointed many fruit powders. This should be a product that is rarely called "ugly" in the Apple series. In view of the fact that the author did not understand art, he hurriedly asked a lot of estheticians in the early morning. This did not make the author suspect that there was a problem with his aesthetic.

HomePod's designers estimated that they were middle-aged people who grew up in the 1980s, because at the first sight of HomePod, they remembered their mother's woolen balls. The young people of the new century saw this antique. Of course, there is a more ugly one. Nylon line ball is also this shape.

Of course, HomePod is ugly, but the performance is not bad, Apple's ultimate pursuit of the user experience is still, and HomePod is by far the first intelligent speaker to return to the nature of speakers. Apple even deployed the microphone array and speaker array at the same time, which is Apple's attitude: not only to pursue the far-field voice interaction experience, but also to pursue the ultimate sound quality to enjoy.

When the boots landed, why use a 6-gear ring array!

HomePod has built-in Apple Siri. This time, Apple has adopted the industry's popular 6-megapixel ring array technology. This microphone array technology is suitable for far-field voice interaction, so that it can satisfy the user's long-distance interaction with the HomePod command via "Hey, Siri". Apple HomePod uses the microphone array technology, which also shows Apple's technological ideas for upgrading Siri from near-field voice interaction to far-field voice interaction.

In the past few years, the most common application for voice interaction was a smart phone represented by Siri. This scenario generally uses a single microphone system. A single microphone system can obtain sound signals that meet voice recognition requirements with low noise, no reverberation, and proximity to the sound source. However, if the sound source is far away from the microphone, and there are a lot of noise, multipath reflections and reverberation in the real environment, the quality of the picked-up signal will be degraded, which will seriously affect the speech recognition rate.

Moreover, the signal received by a single microphone is superimposed by multiple sound sources and environmental noise, and it is difficult to separate the sound sources. This makes it impossible to achieve sound source localization and separation, which is very important because there is still a class of sound superposition that is not noise, but it is also suppressed in speech recognition, which is vocal interference, and speech recognition obviously cannot recognize more than two sounds at the same time. .

Obviously, when the scene of voice interaction transitions to the main scene with smart speakers, smart TVs, robots or cars, the limitations of single microphones are highlighted. In order to solve these limitations of a single microphone, a method of performing voice processing using a microphone array takes place. The microphone array is composed of a set of microphones arranged in a certain geometric structure (commonly used in line and circle) to perform space-time processing on the sound signals collected in different spatial directions to achieve noise suppression, dereverberation, vocal interference suppression, and sound sources. Direction finding, sound source tracking, array gain and other functions, and then improve the quality of speech signal processing, in order to improve the speech recognition rate under real environment.

Judging from the current domestic and foreign market products, Amazon Echo's solution is a 6+1 wheat ring array structure, Amazon Echo Show is an 8 wheat elliptical array structure, Google Home is a 2 microphone structure, the domestic science and technology news The speaker is a 7+1 wheat ring array structure. The current Shengzhi technology product line is the most complete, with 3/4/4+1/6 wheat ring array structure and single wheat, 4 wheat line type, 6 wheat L type, 8 wheat Double L type, 10 wheat distributed arrays and other structures.

In fact, different formations adapt to different scenarios and also consider the price/performance ratio. The more complex the array structure, the higher the cost. For the smart speaker, since the user's habits require 360-degree pickup and orientation, the circular array structure is most suitable. As for the choice of three, four or six, it is determined based on the directional accuracy and the interaction distance. From a certain point of view, it can be understood that the more the number of microphones is more orientated, the farther the speech recognition distance will be. Of course, this There is also a relationship with the specific structure of the formation.

Note that the 2 microphones here are not arrays, and do not have some of the array's features and capabilities. The 2nd mic is most often used to achieve noise reduction on ultra-thin devices such as mobile phones and Bluetooth headsets. In fact, many occasions have been specially designed. A single microphone can replace the 2 bar structure. Since Apple HomePod must be different from Amazon Echo and Google Home, the structure of the six microphones is very sensible and the most cost-effective. This is also the main microphone array of Sonotech. In fact, according to the current technology of Sonic Technologies, 4 microphones are used. The effect will not be too bad, but the voice interaction distance will lose a little.

The basic properties of smart speakers also have to listen to the sound quality!

After all, smart speakers are still the category of speakers. This is a mature category, and they do not understand why many smart speaker manufacturers have to position themselves as robots. In fact, the positioning of robots is actually a disaster for the consumer market, because the robot market is still a market that requires a huge investment in education. Therefore, HomePod chose to return to the essence of the speaker, with a great emphasis on sound quality and listening experience.

The HomePod has great audio technology. At the bottom is a 7-beam array of tweeters that accurately represent acoustics and sound field control. Excessively, in such a small product, Apple actually used a 4-inch low-frequency speaker. Here, it is no longer emphasized that the larger the bass speaker, the better.

Not only that, HomePod also uses a lot of audio algorithms, including automatic bass equalization, dynamic modeling and more. Although the 7-inch figure is small, even if the volume is turned up, the sound quality will not be distorted. HomePod uses the A8 processor chip used in Apple's mobile phones. It also has real-time acoustic modeling, audio beamforming, and multi-channel echo cancellation technology. This makes HomePod the fastest and most sound smart speaker ever. The author believes that because of this reason alone, there will be a lot of fruit powder will be paid.

In addition, Apple also mentioned Spatial awareness technology, in fact, this is not a novel technology, that is, emphasis on the sense of space and immersive, that is, let the music played in different scenes with different sound effects. As the name implies, when the HomePod is in the room, you can adjust the music effect according to the environment of the scene.

Although not new, this is a big step forward because the virtual space sound is extremely dependent on the spatial sound environment. Incidentally, a few words, Dolby tossed so many years of panoramic sound, applied to family-level products has not been able to solve this problem. Millet's ultra-thin TV emphasizes spatial sound, which is reflected in the reflection from the ceiling. However, Dolby obviously cannot adapt to the best sound effects based on the user's home environment.

Of course, HomePod certainly supports multi-room music systems. If you use multiple HomePods, the sound effects will be even better. This is more suitable for young friends who like to meet at home. At present, Sonic Technologies also provides support for multi-room music systems, and also has a technology called "near-wake-up". That is, when multiple voice smart devices exist at the same time, it is the intelligence closest to the user that gives priority to respond to user instructions. equipment.

As for Apple's music ecology and family control, it is no longer emphasized. HomePod's added voiceprint recognition feature is a small highlight, so that Siri will recognize the user's voice is consistent with the user's voice pattern, not only improves the efficiency of use, but also provide security for the user's privacy.

So why is the apple getting uglier?

It seems that not only HomePod, Apple has never launched amazing products since leaving the era of Jobs, and even with closed eyes can guess the shape of the Apple iPhone 8, let alone Mac and iPad series has not changed The design of AirPods, including AirPods, is also a subject of Tucao. This is naturally a result of Cook’s work. The CEO of the supply chain drove the Apple Empire to move forward, but too pragmatic style made Cook lose his control over the combination of technology and art.

Obviously, HomePod is the result of artistic compromise technology, because from the layout point of view, from the bottom to the top is the tweeter array, microphone array, 4 inch woofer and main control board, so listed together, then consider the acoustic structure design From a technical point of view, it is impossible to think of a better shape. But this is always Apple's ah, has the world's most cattle designers and technical personnel, the result was designed a product that did not get rid of technical ideas.

Sometimes it has to be said that letting technology or supply chain managers take charge of product design is also a disaster. Anyway, this is a product fully in line with the aesthetic and style of the technical staff, because technically speaking, this speaker is indeed no problem, but also challenges the technical difficulty, such as the middle of the microphone array technology is very difficult to achieve .

Product defects, how to deal with Apple's play!

However, after all, Apple is Apple, and Apple's understanding and play of products still exceeds the industry's common understanding. First look at the positioning of Apple, high-end this is a must. Amazon Echo is a preconceived product, priced at 179 US dollars, this price is close to the cost of a terribly dead, almost blocked the exit of many products. Forced Google had to go low, bringing the price down to $129, and for this reason it also abandoned the microphone array, sacrificing the far-field voice interaction experience.

In any case, Apple's price is set at $ 349 anyway. This should be a price cut for Apple, but it will also block the way out of the high-end smart speaker market. Secondly, since Apple is positioning high-end, its products must bring high-end experiences to consumers. Therefore, Apple has piled up microphone arrays and speaker array technology, plus the original music and control ecology, to attract a large number of Apple fans. Still a safe strategy. Of course Apple's biggest mistake is that HomePod is a bit ugly, otherwise it will make more manufacturers feel pessimistic.

Judging from the current foreign market, Amaon, Google, and Apple from the low-end to high-end layout for smart speakers, are the same to squeeze the profit margin, this is definitely not a product to make money, but a strategic level product. To put it plainly, the giants did not count on relying on smart speakers to make much money, but could not lose the voice entry. Even if it is not certain whether the future is an entrance, it is better to bet than to miss at least. Moreover, according to the current situation, sound and image are destined to be the two core basic data in the era of artificial intelligence.

This creates a problem. How does Amazon Echo and Google Home react? Amazon is okay, after all, the market occupancy rate is there, and the product line is relatively complete. It is Google instead, and it takes so much energy. Instead, it only serves as a foil to the other two giants. This is good, the most embarrassing is that some domestic manufacturers do overseas markets, such as Lenovo, Lenovo's smart speakers how to face this complex situation? Under such pressure, will there be even more surprising products, such as how Xiaomi should act? This is also what we look forward to most this year.

Why does HomePod wait until the end of the year?

Apple HomePod is expected to wait at least until December to be sold simultaneously in the United States, the United Kingdom, and Australia. Global shipments will wait until later in the year. What is this scenario? A speaker should wait until more than six months. Moreover, according to feedback from friends at the scene, the speakers displayed at the conference should be just a shell, because there are no other functions other than the demo light. Therefore, Apple really has anxious, it must be smart speakers.

It has to be said that this is technically rather embarrassing. Apple definitely guarantees the user experience, but HomePod has added two arrays at once. Which array is not a simple matter. After all, current products are not purely functional products. This is a complete technology chain. For example, the microphone array includes functions such as noise suppression, dereverberation, vocal interference suppression, sound source direction finding, sound source tracking, array gain, model matching, and voice recognition. These are complex technical systems that require careful polishing. Even Apple, it needs enough time to accumulate experience. What the product tests is the details of every place. Therefore, many times, please be kind to those startup companies that work around you day and night.

Why does Apple attach so much importance to a speaker?

With the continuous development of the field of artificial intelligence, people began to pursue more free speech interaction methods, and the advantages of far-field speech interaction gradually became apparent. In fact, prior to Echo's appearance, voice interactive products had always been solved near-field problems. This is a typical case of deliberately avoiding scenarios due to technical limitations because near-field speech interaction requires humans to adapt to the machine.

However, the voice interaction between humans has always been a certain distance, so now requires the machine to adapt to humans. This can be said to be a great advancement in computer technology and one of the core elements of artificial intelligence.

Of course, this is not a problem unique to the acoustics field. When the camera and radar are installed in the car and the GPS is installed in the bicycle, the technical challenges brought by the scene change will be highlighted because the technical support needed for the real scene is not simply upgraded. However, it is disruptive innovation. This is the main reason why giant companies have entered this field one after another. No one wants to be eliminated in the process of upgrading technology.

In fact, when the fusion data acquired by the machine is sufficient to cover one tenth of the human population, humans do need to speak, look, or think about the machine for many times. But at this time, we do not know. What new business models will be generated in the end, after all, from our perspective, the advertising model is certainly not the best business model in the era of artificial intelligence.

From Echo Amazon's best-selling, everyone gradually turned their attention to smart speakers. Google launched Google Home, and Microsoft teamed up with Harman Kardon and Hewlett-Packard to launch smart speakers equipped with Cortana. Although Apple used the smart headset Airpod to seize the entrance to the voice market, however, with Amazon Alexa through Echo in the smart home market, it seems that it gradually began to swallow Apple's market share in the smart home, and gradually stabilize its voice interactive portal, Alexa It seems to be the new generation of "Android" or "OS." It seems that Apple's launch of HomePod is indeed imperative.

summary

Even in the era of Jobs, Apple’s product release will be repeated many times, but Apple's sales are the best response, at least, from the performance and price of Apple HomePod, HomePod sales will not be bad, this from Airpods can be analogized. In particular, Apple actually imitated the Xiaomi line, and compared the price quite seriously. It finally gave the price of a competition Xiaomi. Looking forward to the major domestic giants, how to deal with Apple's strategy to play it?