Research on indoor sound reinforcement system based on DSP

1. Introduction

　　For traditional indoor sound reinforcement systems, the mode is generally that the microphone picks up the sound, then sends the signal to the power amplifier for amplification, and then sends it to the speaker for playback. This sound reinforcement system does not do any processing on the noise, which leads to the deterioration of the indoor sound field characteristics and even affects the audience's ability to accurately hear the expected signal. This paper presents an improved sound reinforcement system design scheme, which adopts a microphone array and uses digital signal processing technology to suppress useless signals as much as possible without affecting the real-time performance of the signal.

　　2. Indoor sound reinforcement system based on DSP

　　2.1 Indoor sound field characteristics

　　For indoor sound reinforcement systems, the sound field characteristics are relatively complex. The signals reaching the microphone may include desired signals, reverberation signals, interference signals, noise signals, and even the playback signals of the speakers (related to the position of the indoor audio and its directivity). This sound field characteristic can be represented by Figure 1.

　　Among these signals, except for the desired signal (what we need), all other signals will affect the clarity of the sound signal, so these signals can be considered as noise, especially the playback signal of the sound system, which may cause howling through acoustic feedback, seriously affecting the stability of the indoor sound field. Therefore, how to eliminate or reduce these unnecessary signals as much as possible, maintain the stability of the indoor sound field and the clarity of the language signal, is the first problem that the indoor sound reinforcement system should consider and solve.

　　2.2 Signal model in this design

　　The sound reinforcement system design proposed in this paper adopts modern array signal processing technology, uses sensor arrays, combines digital signal processing technology and corresponding algorithms, so that the signal output by the array has a larger signal-to-noise ratio for the desired signal, and has a larger attenuation for useless signals other than the desired signal (the ideal result is to attenuate to zero). The processing diagram is shown in Figure 2:

       3 Hardware Design of This Solution

　　3.1 Design Block Diagram In this sound reinforcement system, a microphone array and a digital signal processor are used to suppress useless signals according to relevant algorithms. The block diagram is shown in Figure 3:

　　In this design, the signal sensed by the microphone is first low-pass filtered. Since the expected signals of the indoor sound reinforcement system are generally voice signals, after low-pass filtering, the frequency of the output signal is limited to within 3400Hz. Then, each signal is converted into an analog-to-digital signal to facilitate the subsequent processing by the digital signal processor. After weighted summation of each signal, the output signal is sent to the subsequent power amplifier for amplification and then sent to the speaker for playback. The key here is the automatic update of the weights. As long as an effective algorithm is used, the DSP will automatically correct the weights so that the expected signal can be output correctly and the useless signal can be suppressed as much as possible.

3.2 Device Selection

　　In this scheme, the microphone can be a general omnidirectional microphone, and the low-pass filter can be a common low-pass filter. The purpose of adding the filter here is to limit the frequency of the sound signal to within 3400Hz. The analog audio signal after filtering needs to be digitized, that is, A/D conversion, that is, the analog signal is converted into a digital signal through sampling, quantization, and encoding. In this process, due to the quantization error (noise) in quantization, signal loss is inevitable, especially for small signals, which may be completely lost. Therefore, if the dynamic range of the A/D conversion device is insufficient, many useful small signals will be quantized to zero. Therefore, under the condition of meeting the sampling frequency, an A/D converter with higher sampling accuracy should be selected to make the quantization distortion as small as possible (this is also a principle of device selection). Of course, in this scheme, since the frequency of the analog signal directly processed is low, AD7870 and AD7870A on the market are good A/D converters with an accuracy of 12bit.

　　After analog-to-digital conversion, each digital signal is sent to the digital signal processing module for processing. For the sound reinforcement system, under the premise of effectively suppressing various useless signals (depending on the effectiveness of the algorithm), the real-time requirements are very high. This places certain requirements on the selection of digital signal processors. For digital signal processing devices, the speed, accuracy, and memory size of the operation should be considered when selecting.

　　In this design, the DSP chip selected is Ti's TMS320VC5509, which integrates a C54x core, 128KB 16-bit on-chip RAM memory, and a maximum of 8MB 16-bit external storage space. Its main features are [4]:

　　CPU: two multiplication-accumulation units (MACs); a 40-bit arithmetic logic unit and a 16-bit arithmetic logic unit; multi-bus structure, etc.

　　Memory: 128KB 16-bit on-chip RAM memory; 8MB 16-bit external storage space, etc.

　　On-chip peripherals: 2 20-bit timers; 6-channel direct memory access controller (DMA), etc.

　　3.3 Simulation results

　　This simulation assumes that the microphone array consists of 4 microphones in a linear array, uses the MUSIC algorithm [5][6], the distance between adjacent microphones is 8cm, and the frequency is 2kHz. The simulation is divided into two cases. Figure 4 assumes that there are two signals we need, and the input signal-to-noise ratio is 34dB (which can be achieved in actual sound reinforcement systems). The simulation results when the incident angles of the two desired signals are 30 degrees and 70 degrees respectively. It can be seen that in the two useful signal directions, the microphone array has a large power spectrum output, while in the remaining directions (corresponding to useless signals), the power spectrum is very small [4]. (In the figure, the horizontal axis represents the incident angle, and the vertical axis represents the array output power spectrum density.)

　　FIG5 is a simulation result when the input signal-to-noise ratio is 25 dB and the incident angle of the expected signal is 15 degrees.

　　From the simulation results, for useful signals, there are larger output results (which can also be understood as a larger signal-to-noise ratio), while for useless signals, they are all attenuated to very small results, which shows the effectiveness of the algorithm.

　　4 Conclusion

　　There are many ways to improve the performance of indoor sound reinforcement systems, such as architectural acoustics, electroacoustics, etc. Theoretically, the sound reinforcement system based on DSP technology proposed in this article can indeed improve the performance. At the same time, the system is not very complicated. It only needs to add an array processing part (which can be integrated into the power amplifier) ​​to the sound reinforcement system, and then go through the opposite processing process. Of course, how to improve the system performance more effectively still requires the continued efforts of scientific researchers. The author's innovation in this article: modern signal processing technology is applied to indoor sound reinforcement systems, and an array signal processing part is added to the sound reinforcement system. Using efficient and effective algorithms, combined with digital signal processing technology, and using digital signal processing devices, the useless signals in the indoor sound field are well suppressed.