How is the global microphone technology market, how to choose far-field voice interactive microphone?

Electronic enthusiasts at 8 o'clock in the morning: the fierceness of the intelligent voice interaction market has gradually radiated to the suppliers of the industrial chain, and the most direct benefit is the microphone as a sound sensing device. In particular, the rise of microphone arrays can double the sales of microphone manufacturers in the future. Prior to this, due to the decline in the growth rate of smartphones and tablets, the shares of Lou, Goer and AAC successively hit a trough around mid-2016.

In 2017, the situation of voice interaction tends to be clear, and the number of investment institutions that focus on smart voice interaction in the capital market has obviously increased. Then, it is necessary to have a deep understanding of the microphone industry and technology trends.

As a popular name in the industry, the microphone is the transliteration name of English Microphone. The domestic name is a little messy. Sometimes it is also simply called the microphone. It is also called the microphone and the pickup in Hong Kong and Taiwan. The Chinese academic name of the microphone is officially translated as a microphone. This is a transducer that converts sound into electronic signals, that is, converts sound signals into electrical signals. This is actually exactly the same as the principle of photoelectric conversion.

The microphones in the consumer market are basically scalar microphones, which means that only a single physical quantity information, sound pressure, can be collected. What is sound pressure? Sound pressure refers to the amount of pressure change caused by vibration when sound waves pass through a medium. It can also be understood as the amplitude or intensity of sound. The sound pressure is usually indicated by the letter "p" and the unit is Pascal (symbol Pa). The Pascal unit of sound pressure is inconvenient to memorize (such as 20x10-6Pa ~ 20Pa), and the effective sound pressure is generally measured on a logarithmic scale relative to a reference value, that is, the sound pressure level, whose unit is decibel (symbol dB ).

The human hearing threshold for 1KHz sound is 20 x 10-6Pa, which is usually used as the reference value of the sound pressure level. This may be obscure. Let ’s make a simple analogy: the sound pressure of human breath is about 60x10-6Pa, the sound pressure level is about 10dB, the sound pressure of the rocket launch is about 4000Pa, the sound pressure level is about 165dB, the sound of the flash bomb The pressure exceeds 10,000 Pa, and the sound pressure level is about 175 dB.

Analysis of these obscure numbers will naturally lead to the parameters of the microphone. In fact, our concern is the ability of the microphone to restore the true sound. This is a difficult challenge, because the conversion of sound to electricity means distortion.

In order to describe the performance of the microphone, several performance indicators are very critical, including sensitivity, directivity, frequency response, impedance, dynamic range, signal-to-noise ratio, maximum sound pressure level (or AOP, acoustic overload point), consistent Sex and so on. These indicators are actually easy to understand and determine the performance of the microphone, and each indicator is very critical and indispensable. Of course, these indicators are really simple compared to the TS parameters of the speaker.

The microphone is a typical sensor industry. Its technological iteration is very rapid, and its appearance has changed a lot. It is estimated that it is difficult for many people to find the microphone accurately from the microphone array below.

Now the microphone array mainly uses digital MEMS microphones, whose longest size is only 3.76MM. MEMS microphones are also used as sensors in mobile phones. Generally, mobile phones have at least two or more such microphones. The MEMS microphone is actually only an improvement in technology, and its principle still belongs to the condenser microphone. The direct PK with the MEMS microphone is the electret microphone.

These two microphones are the main force in the field of consumer electronics. For example, the development boards of Amazon's Echo and Acoustic Technology are mainly MEMS microphones, and the development boards of SciTech are mainly electret microphones. From the performance indicators, there is no substantial difference between the two microphones. The performance indicators of the electret microphones are still higher. Therefore, the single microphone series of Shengzhi Technology is also an electret microphone. However, the advantage of MEMS microphones is the consistency comparison. Well, it is more suitable for the microphone array for far-field voice interaction.

There is another shape of condenser microphones, which are the professional microphones commonly used in recording studios. This is essentially different from the microphones used by our KTV singing. KTV microphones are generally dynamic coil microphones. The performance of this microphone is not as good as condenser microphones. The advantage is that it is suitable for vocal recording. The disadvantage is low sensitivity. This is an advantage in KTV because it can effectively avoid howling in the KTV environment. Of course, there are band microphones and carbon microphones. These two microphones are no longer common, especially carbon microphones, which were mainly used in old-fashioned phones, but are now basically eliminated. These types of microphones can be compared from the picture, and sometimes they will sigh that technology has always been subversive.

Of course, the new type of microphones also include piezoelectric microphones, fiber optic microphones, laser microphones, etc. You can even use TV screens or speakers (speakers) as microphones. In addition, the defense field is also researching and deploying more complicated microphones such as vector microphones. In fact, the development trajectory of sound is similar to that of radar. The microphone array is also a stage of current technological development. The iteration of sensing technology is an important reason for the iteration of technology and market.

There is no doubt that in the semiconductor ecosystem, MEMS is playing an increasingly important role, and its application range includes consumer electronics, automotive industry, industrial control and even biomedicine, aerospace and other fields, and is still expanding rapidly. With the growth of the consumer electronics market, consumer applications have become the main driving force of the MEMS sensor market. In 2014, the MEMS sensor market reached US $ 13 billion, and the largest consumer application reached US $ 5.9 billion. It is expected to exceed US $ 25 billion by 2019, with a compound annual growth rate of approximately 11.2%.

We compared the latest industry research data. From 2013 to the present, the market pattern has not changed as expected. Instead, MEMS manufacturers have entered a period of low-speed development in recent years. Fortunately, the voice interaction market has been hot this year. Brought imagination to the MEMS microphone market. The following is an overview of the market with data published on the Internet. It has been publicly reported for 3 years. I believe that many doubts have also disappeared. These data do not represent the real situation, especially the current market status, but they can be used as an important reference.

According to IHS statistics, the US company Louth Electronics was the world's largest supplier of packaged MEMS microphones (direct printed circuit board assembly) in 2013; the German company Infineon is a MEMS microphone bare crystal (supplied to MEMS microphones Manufacturer). Lou's revenue accounts for 59% of the packaged MEMS microphone market, while Infineon's shipments account for 78% of the MEMS microphone die market.

Revenue ranking of global packaged MEMS microphone suppliers (unit: million US dollars)

Global MEMS microphone bare chip supplier shipment list

In the packaged MEMS microphone market, the second-ranked supplier is AAC, with a revenue market share of 13%; the third-ranked company is Goertek, with a revenue market share of 7%. AAC and Goer are both Chinese manufacturers, and their businesses are heavily dependent on Apple and Samsung; AAC supplies one of the three high-performance MEMS microphones for iPhone5 and also stationed in iPhone5S; Goer is the main MEMS microphone for iPhone earphone supplier. AAC's revenue grew by 8% in 2013, while GoerTek started to supply high-performance microphones for mobile phones, and its revenue growth rate in 2013 was as high as 35%. Of course, after these three years of market changes, the market share of these three companies has actually changed quietly. Lou's continued to strengthen its leading position in the industry.

The world ’s fourth-largest packaged MEMS microphone supplier is South Korea ’s BSE, whose performance comes from the big customer, Samsung, which has Apple ’s largest rival; BSE ’s 2013 revenue compared with 2012 has a growth rate of more than 250%. The fifth largest manufacturer after BSE is European STMicroelectronics (ST). The company ’s MEMS microphone business continued to expand because it won the iPad design.

The world's top 50 MEMS manufacturers

There are also some microphone manufacturers in China, including Taiwan. These manufacturers mainly rely on price to win, and technical barriers have not been established too high. There are also some start-up companies at home and abroad, such as Vesper MEMS in the United States, which uses piezoelectric microphones as the main technical characteristics. However, these technologies are themselves mastered by first-line manufacturers, and only the market factors are considered without production line deployment. So this type of startup company is still under great pressure when it enters mass production, because this industry really needs heavy assets and heavy investment.

The concept of far-field voice interaction is actually relative. We know that voice interaction involves a very complex technology chain, including core technologies such as acoustic processing, speech recognition, semantic understanding, and speech synthesis. Acoustic processing is mainly to simulate human ears, to ensure that the machine can listen to the human voice in a real environment, speech recognition is to translate the human voice into text, and semantic understanding is to analyze the meaning of these words. The machine is synthesized by speech synthesis. The text to be expressed is translated into speech. Although these four technologies have developed independently, they cannot actually be separated. At the same time, with the cooperation of other technologies, a complete chain of voice interaction can be formed.

Near-field speech recognition, represented by Siri, has been developed for more than 60 years, especially after deep learning has been substantially improved since 2009, but as Zuckerberg said, when real products land, we find users What is really needed is far-field speech recognition similar to that advocated by Echo. Obviously, this is a new technical field, because the problem of expanding the pickup distance is not only the attenuation of the voice signal, but also brings a complex real environment and complex user habits.

Near-field speech recognition requirements must be low noise, no reverberation, and very close to the sound source. For example, users must always speak into the mobile phone to obtain sound signals that meet the requirements of near-field speech recognition, and also require users to meet standard pronunciation , Its recognition rate is likely to reach more than 95%. However, if the sound source is far away, and there is a lot of noise, multipath reflection and reverberation in the real environment, the quality of the picked-up signal is reduced, which will seriously affect the speech recognition rate. Similarly, our human performance in complex far-field environments is not as good as whispering.

Generally, in the near-field speech recognition engine in the far-field environment, if there is no support for acoustic processing, such as the adaptation of the microphone array technology, the actual scene recognition rate is actually less than 60%. Moreover, since real scenes always have multiple sound sources and environmental noise superimposed, such as scenes where peripheral noise interferes and multiple people speak at the same time, this makes voice recognition more difficult. Because the current speech recognition engines are all single-person recognition modes, they cannot handle the problem of multi-person recognition at the same time.

The microphone array is currently the main way to solve the above problems, but the microphone array also has many shortcomings, one of which is higher hardware requirements, which includes microphones and chip devices. Therefore, how to select a microphone in the microphone array is also a very careful and serious task. So, how to choose a microphone for intelligent voice interaction?

First of all, we must have a clear definition of the product before selection, such as product sales, product life, product scenarios, etc. This is very important. Even if the performance indicators are completely the same, the cost, production process, supply capacity and other factors of the microphone supplier need to be considered, because the sensor component manufacturer is ultimately a production process problem. For example, many domestic microphone manufacturers actually buy foreign products directly. Infineon's MEMS microphone solution is directly packaged and does not have technical research and development capabilities. Its production process and production scale are the main differences between many domestic manufacturers.

Secondly, the type selection should focus on evaluating the performance indicators of the microphone, which include sensitivity, directivity, frequency response, impedance, dynamic range, signal-to-noise ratio, maximum sound pressure level (or AOP, acoustic overload point), and consistency. Among them, directivity, signal-to-noise ratio, AOP and consistency are the most important indicators considered by the microphone array. In general, there are the following types of directivity, and there is no substantial difference between major manufacturers. The signal-to-noise ratio (SNR) and AOP are not too different for the three giants Lou, Goer and AAC. For example, the SNR is basically above 65dB, and the AOP is also greater than 120dB. The microphone quality above this parameter It is currently more reliable and mature, and is also the main supply model of the first-line big manufacturers. Consistency is an indicator that needs special consideration for microphone arrays, because MEMS microphones need to be welded, which cannot place too harsh requirements on welding machine manufacturers. After welding, can the microphone still guarantee a certain performance? This is for microphone manufacturers. A huge test. Therefore, don't believe in the technical indicators and commitments provided by microphone manufacturers. You need to verify their factories and production workshops and management.

The final selection must actually check the quality of the microphone, that is, the yield rate. From the experience of Shengzhi Technology, when mass production, sometimes the batch is different, resulting in a large difference in microphones, especially some small and medium microphones Suppliers. Moreover, it is particularly troublesome that the replacement cost of MEMS is also very large. Once the patch is soldered, if there is a problem, it will often lead to the scrapping of batches of microphones. Of course, this only needs to be considered when the volume is large. When the volume is small, this issue is not too critical.

But it has to be said that the market competition of MEMS has always been very fierce. It was also mentioned earlier that giants such as Lou, Goer, AAC, STM, and Invense have been born in this market, and occupy more than 80% of the market. Although the market is gradually transitioning from mobile phones and tablets to IOT devices, the current stock market is still mobile phones and tablets, and the amount of IOT is not too large. So this raises two question marks: one is whether this market will be a monopoly or a giant decline in the future? Second, is there any business opportunity in this market?

Although the emerging voice interaction market is a strong demand for microphone devices, will this change in scenario lead to changes in the market of giants? It is really difficult to say in the future, only a few thoughts can be given for reference: one is voice The strong demand in the interactive market is the microphone array, not the components of the microphone, because for the microphone manufacturer, in addition to adding more customers, its customers have not changed substantially and there is no higher technical demand. From this point of view, the reasons for the changes in the existing MEMS market pattern are not sufficient.

Secondly, from a technical perspective, all technology manufacturers need to pay attention to technology upgrade iterations, such as the need for low power consumption, high integration, and vectorized microphones. If these first-tier manufacturers do not track the progress of technology, when the technology is subverted It is likely to be replaced at some time, and the suppliers of electret microphones are likely to face this serious problem in the short term. Of course, the situation of some companies is always awkward, especially when some manufacturers encounter microphone technology and market challenges, they choose to retreat and transform, and the transformation is often made into a transition, which is a huge challenge for sensor manufacturers.

So, there are no new entrepreneurial opportunities in the microphone industry? From the perspective of equal opportunities, there are many entrepreneurial opportunities in any industry. The core lies in how to grasp and use. This is somewhat similar to the chip industry. The industry attributes of the microphone determine that it must be a heavy asset model, including the scale of investment in equipment and production lines. In addition, cost control, production line management and processing technology are also its core elements. These core Virtually every item is a huge investment.

In addition, there are ideal manufacturers who should stick to the technology-driven route. For example, AAC and GoerTek initially bought Infineon ’s MEMS and ASIC chips, and then packaged and sold them to customers. Using the acquisition, the two companies have gradually mastered MEMS microphone technology. So for other manufacturers, is it a very painful process whether this path can be copied or surpassed, or simply transformed.

The future market space of MEMS microphones is bound to be larger, and its technical requirements will be higher and higher. Under the situation that giants are already established, how can emerging startups seize the opportunity to break through or share cakes, this is a painful test. Whether to withstand the pressure to face the competition directly, or to retreat and transform into other markets, this is a realistic question often faced by all entrepreneurs.

Disclaimer: The works reproduced on the website of Electronic Audiophile all indicate the source as much as possible, and all rights of the owner of the work will not be transferred due to the reprinting of this site. If the author does not agree to reprint, please inform this site to delete or correct it. The reprinted works may be changed in the title or content.

Shenzhen Kate Technology Co., Ltd. , https://www.katevape.com