Adjusting system parameters dynamically to improve

  • Detail

Dynamic adjustment of system parameters to improve echo and noise cancellation performance

the voice quality of mobile handheld terminals, wireless networks, hands-free devices and other mobile communication systems is the key factor to establish consumer preferences. Echo and noise are inherent defects of wireless communication. We need signal processing technology to solve the problem of voice quality and ensure that high-quality audio output can be accepted by the market. The traditional method is to use independent echo and noise cancellation modules on the near end or transmission path. This method performs well when the surrounding conditions remain unchanged. However, if the surrounding environment changes, such as door opening or large noise, the audio system will be difficult to adapt to the changes, and the audio performance will decline

the new method integrates echo cancellation, noise suppression and other sound enhancement technologies, and can quickly and dynamically adjust system parameters according to environmental changes. In most cases, we can complete the adjustment before the consumer finds the sound quality problem. The development of eco-environmental materials and their green production technology has attracted more and more attention. Similarly, this new method has achieved a higher degree of integration and can solve the problem of large noise and echo, so that it can realize full duplex voice calls that are considered very natural by the industry inspectors

the great progress in echo and noise cancellation technology came at a very timely time, because many states in the United States have enacted relevant laws and regulations to prohibit drivers from holding mobile phones while driving in full or in part. Most countries in Europe and many other countries around the world have also had relevant regulations. The emergence of the above regulations has further increased the demand for hands-free technology, and requires that the noise and echo can be effectively eliminated in the internal environment of the vehicle, which is also the biggest design challenge of the hands-free system. Designers need easy-to-use software and hardware to provide hands-free audio products with the same sound quality as traditional handheld products, so as to meet the needs of users

echo sources in wireless communication

there are two major sources of echo in wireless systems: electrical echo and acoustic echo. If the poor design results in the direct coupling of the speaker signal to the loudspeaker signal, an electrical echo will occur. The best solution to this problem is to do a good job in design

the problem that poses a more serious challenge to us is acoustic echo. If the amplified loudspeaker signal is echoed through the loudspeaker, an acoustic echo will appear. It is quite difficult to eliminate this kind of echo. We must consider several factors. The amplified speaker sound will be reflected on multiple channels at different times. This indirect echo obviously lags behind the original signal, because the propagation speed of sound in the air is only 300 m/s, and the echo reflection will be distorted due to the increased complexity of mechanical vibration

half duplex switching technology

Figure 1: half duplex solution

the most basic way to solve the echo problem is to disable the near end voice channel when the far end voice is detected. Although this can eliminate the acoustic echo, only one person is allowed to speak at a time. For example, as for the traditional interphone, if you press the call button, you can't hear what other people say. Therefore, the two-way radio call rules require that the speaker must clearly indicate that he has finished speaking. Later, a new technology replaced the call button with voice activity detector (VAD), which can automatically turn on/off the near end voice channel when the far end voice is detected. At the initial stage of mobile communication, we can still accept this limited technology. However, as users become accustomed to full duplex wired calls, they will no longer accept this limited one-way call technology in the future. Because full duplex wired communication technology enables them to communicate freely, express their ideas, agree or disagree with each other's views, and stop at any time without worrying about the sudden failure of the microphone

traditional echo cancellation technology

Figure 2: traditional echo cancellation technology

next, we will tell why almost all hands-free devices and those with loudspeakers provide some kind of echo cancellation technology. At present, almost all devices eliminate the echo by monitoring the remote signal and then removing the remote signal from the received signal. If the echo quantity is known and remains unchanged, the above method can easily realize echo cancellation. However, in fact, the echo amplitude and time depend on the environment in which the wireless equipment is used, and this environment often changes. Therefore, the traditional echo cancellation technology needs to continuously monitor the near end and far end signals. The acoustic echo canceller algorithm uses the reference signal of the near end loudspeaker to estimate the echo channel, and removes the echo from the near end loudspeaker signal

the design and adjustment of adaptive filter is the decisive factor of echo cancellation performance. The filter usually uses the known characteristics of the audio signal to calculate the echo estimation, and adjusts the parameters of the filter to minimize the error. We usually use the normalized least mean square (NLMS) algorithm to update the filter coefficients to eliminate the echo. The algorithm can minimize the mean square error of the canceller, which is residual echo. Adaptions are usually normalized according to the signal power to be independent of the signal level

in most cases, we can perform the above calculation with sufficient accuracy to reduce the perceptible echo. The problem is that whether the algorithm can work depends on the stability of the echo path between the loudspeaker and the loudspeaker. As long as there are objects nearby that hinder sound transmission (for example, put your hands on the desktop, touch the keypad, and cover the paper on the speaker), or when the distance from the microphone to the speaker changes (put the wired microphone back to its original position), the echo path will change. When the path changes, the algorithm will be adjusted according to the new echo path, which will cause delay. In the adaptive delay process, the echo will be transmitted on the near end signal path

when designing an echo canceller, it is very important to understand the working environment of this device. Are the loudspeakers and speakers in a fixed position? Will the position change affect the normal operation? What is the longest echo path allowed by the operating environment of the device? How much noise is expected? Does the noise change (e.g. in a car environment)? How loud should the device be? What is the echo return loss between the speaker and the amplifier? How much should the voice volume of the near end caller compare to the echo at the microphone end? Only by answering the above questions can we design the best traditional echo canceller that can be adjusted according to the known environment. However, when the environment changes, the filter coefficients are still adapting to the new echo channel, and we have heard the echo. According to different initial parameter settings, the adaptation process takes 5 to 10 seconds

in addition to the problem that echo affects the quality of near end signal, background noise can also cause adverse effects. The solution to this problem is to use noise canceller. A typical noise canceller works independently of the echo canceller, and any interference problem can be ignored. Unlike the echo canceller, the noise canceller has no reference signal as the basis. It must estimate the noise and remove it from the speaker signal, or it can only estimate the voice. In either case, the noise should be aimed to maximize performance. By combining the noise canceller with the AEC control signal, we can achieve a more accurate voice activity detection environment and improve the integration effect. Without the above interaction, the system may mistakenly eliminate the speech signal as noise

the new method improves the integration degree

Figure 3: the new method integrates echo and noise cancellation with other audio processing technologies

in order to solve the limitations of traditional technologies, we have developed a new method that can improve the quality of wireless audio (as shown in Figure 3). The basic difference between the new and old methods is that the new method integrates echo cancellation and noise cancellation with other audio signal processing functions, and is uniformly controlled by the new full duplex control module. This method uses the same core NLMS algorithm, but has some special features, which can not only give full play to the unique breadth of system technology advantages of this integrated method, but also dynamically adjust system parameters to accelerate the re integration of NLMS

full duplex control technology is the key to improve the performance of the new method. By combining the audio part of wireless communication equipment with the latest digital signal processing technology, nonlinear control algorithms can be used to adjust sudden environmental changes, such as the door closing sound in the background or the sudden gesture or action of the user's hand. Because different control algorithms are optimized at the same time under the main controller, the sound quality is further improved. A US trader said that finally, due to the adoption of a more powerful signal processing architecture, we can add new functions, such as filling the background with comfortable noise generated naturally to compensate for the change of noise background and avoid noise pumping

it is very difficult for previous generation DSP to integrate the system processing technology of all key components of near end and far end audio path, so as to optimize the signal quality at both ends of the call. The recently introduced DSP has achieved a proper balance between performance and advanced on-chip memory capacity. The complexity of its algorithm and the integration of audio processing are enough to meet the requirements of rapid optimization of different audio components, and help to achieve the best wireless voice quality

working principle of the new method

the new method uses the whole system to understand the current working environment and dynamically adjust the system parameters to obtain the best performance. Analysis and parameter adjustment are the tasks of integrated full duplex control. Full duplex control technology can evaluate the near end and far end signals. First, determine whether the signal is currently in working state, and then evaluate the signal quality from different angles. According to the above information, the full duplex control mechanism will comprehensively and dynamically adjust each module to improve the quality of near end and far end signals

the full duplex control mechanism on the near end signal channel controls the parameters of the nonlinear processor, echo canceller and noise canceller to reduce echo and noise. The full duplex control mechanism on the remote signal path controls the dynamic processing mechanism, adjusts the audio signal, and improves the volume output while reducing the speaker nonlinearity. Graphic equalizer and sound quality enhancement technology are used in both signal paths. The graphic equalizer is used to adjust the transmitter (speaker and amplifier) and the frequency characteristics of the audio signal. Sound quality enhancement technology is used to adjust the sound quality to achieve the best speech intelligibility

the feature of using this system according to the equipment structure and use situation technology is that the full duplex control technology uses the environmental information learned by the system to achieve higher volume and lower echo, and can quickly adapt to the changing environment

design a new audio processing system

the integration of the new audio processing system has been greatly improved, which puts forward a series of design challenges to us. First of all, we should find an appropriate DSP, which can provide the required high performance for the new design and an appropriate programming environment to support the design with much higher complexity than the traditional echo and noise cancellation technology, so as to shorten the design cycle of the mobile communication system

for example, Texas Instruments

Copyright © 2011 JIN SHI