About mobile phone recording and noise reduction - 上海419贵族宝贝-上海后花园1314-上海品茶网-阿拉爱上海

The writer of this article is He Shunyi, an engineer of Ketong Core City.

It must be that we all have the experience that when we receive a call from a friend from a railway station, subway, meeting place, KTV and other occasions, it is sometimes hard to hear clearly, but sometimes it is very clear. Why is this?

Usually we think that the other party's signal is unstable, so the call quality is good or bad. In fact, whether you can hear the other person's speech clearly in this environment depends on the quality of the other person's mobile phone recording and noise reduction functions. At the same time, this is an important difference between high-end mobile phones and ordinary mobile phones.

The difference of any function, in the final analysis, is the difference between hardware and software. In this article, the author will spend a long time to share with netizens the principles of mobile phone recording and noise reduction; Required hardware and algorithm; And different hardware, algorithms, and experience. I hope it can help you.

Recording process and hardware

First, let's talk about why we should emphasize the recording function of mobile phones.

Quite simply, mobile phones are used to make calls. In the process of communication, the voice of the speaker should be recorded first, and then the listener can hear it. Therefore, the recording function is fundamental and important for calls.

To put it simply, the recording process of mobile phones needs to go through three stages and two links. The three links are: "sound - analog electrical signal - digital electrical signal". The two links are "microphone" and "ADC (analog digital converter/digital analog converter)". The microphone is responsible for converting "sound" into "analog electrical signal", and ADC is responsible for converting "analog electrical signal" into "digital electrical signal". Therefore, the quality of the microphone and ADC directly determines the quality of the recording function.

The microphone is familiar to everyone, so I will not repeat it here, but mainly talk about ADC.

How to measure the quality of an ADC? To put it simply, look at two parameters: sampling speed and quantized digits. What are the sampling speed and quantization bits? It can be understood that the sampling speed represents the speed, and the quantized digits represent the precision. The larger the two values, the better.

Then, how to know the "sampling speed" and "quantized digits" of the ADC in the mobile phone? There are ways:

First download a free APP called "RecForg". After installation and running, find the "Settings" menu and enter the rear interface as shown in the figure ▼

In the above figure, there are two red boxes: "Sample Rate" and "Audio Format". These two submenus correspond to "Sampling Speed" and "Quantization Bits" of ADC respectively.

On the author's mobile phone, click "Sampling Rate" to enter the following interface ▼

It can be seen that there are three gears that are not gray selectable: 12kHz, 24kHz and 48kHz. All other gears can be selected. This shows that the "sampling rate" of the author's mobile phone ADC has five gears, and the maximum is 44kHz. At the same time, the author also tested his friend's Xiaomi 2 and found that its highest sampling rate was 48 kHz. This shows that the ADC used by Xiaomi 2 is one level higher than that of the author's mobile phone.

In the "Settings" menu interface, click the "Audio Format" submenu to enter, and the following figure ▼ will be shown

The figure above shows that the "quantized digits" of the ADC of my mobile phone is 16.

Is it easy? It should be noted that the author found that APP "RecForg" can only be found on Android mobile phones, but not in IOS. If you want to check the parameters of the iPhone ADC, you can try to find some similar recording software and try your luck. Maybe you will find something.

Principle and algorithm of noise reduction

In the "Recording Process and Hardware" section, we talked about the hardware required for recording and the impact of hardware performance on recording quality.

In a quiet environment, the software has little impact on the call. However, the mobile phone is a mobile communication device, and the call scene is uncertain, which may be in a noisy environment. In this case, noise reduction algorithm is very important for call quality.

What is noise reduction? To put it simply, the algorithm separates voice and noise from the received voice, strengthens voice and suppresses noise, so as to improve call quality. The reason is very simple, but the algorithm is very complex in terms of specific implementation. Generally, mobile phone companies will not do their own noise reduction algorithm, but adopt the scheme of relevant professional companies.

When it comes to noise reduction, we have to mention the Audience Company. This is a global leading company specializing in mobile communication audio technology. In general, it is a company that does audio noise reduction algorithms. Apple, Samsung HTC、Google、LG、 Huawei, Sharp, Meizu, Xiaomi, etc. are all audience customers. If you want to list the models that use audio chips, it will be a very long list.

So, what will be the difference between different noise reduction algorithms to reflect the user experience?

We can conduct two kinds of tests in noisy environment. （1） In a noisy environment, what is the maximum distance between the speaker and the mobile phone when the other party can hear clearly by using the "hands-free call"? （2） What is the maximum distance that a mobile phone can recognize speech in a noisy environment? The author has tested some high-end mobile phones and ordinary mobile phones, and the results are quite different. If you are interested, you can give it a try.

The above is the noise reduction performance from the perspective of experience. In addition to subjective feelings, can there be an objective and intuitive presentation? The answer is yes. See the figure ▼ below

The above figure shows the frequency response curve of the iPhone 4s and Xiaomi 2 playback channels. It can be clearly seen that the iPhone 4s has done corresponding noise reduction processing in both low frequency (<80Hz) and high frequency (>1.4Hz) bands, and only the voice band and voice band are reserved. Xiaomi 2 only has noise reduction in the high frequency band. This also shows that compared with the iPhone 4S, Xiaomi still has some room for improvement in some details.

As for the principle of the above test process, the software used, the test steps, the meaning of the test items, etc., limited to space, the author will detail in other articles.

Learn more about new cool devices, stay tuned