ARM Cortex-M audio performance analysis

In recent years, portable music players have shined and become popular. There are many reasons for this, such as the massive amount of compressed music content, the cost of data storage is declining, the operation is simple and the online music content of different popular elements is easily accessed. Consumers are increasingly picky about next-generation music players, expecting longer audio playback times and a richer listening experience.

This resulted in the Cortex-M series, which is a new generation of ARM low-cost microprocessors designed with low power consumption. The Cortex-M3 core and the recently released Cortex-M4 core are based on the three-stage pipeline of the Harvard architecture and use the Thumb-2 instruction set architecture (ISA), with lower memory requirements. But can these MCUs be competent for audio processing tasks? Can they create a better listening experience and meet people's expectations?

In order to analyze whether these processors are suitable for processing audio, first of all, we take MP3 decoders and equalizers as an example to roughly understand the implementation of audio processing modules commonly used in audio components such as audio decoders and post-processing, and then implement the processing of these modules according to high efficiency According to the requirements of the device instruction set architecture, the modules are classified.

We focus on the advantages of the Cortex-M3 and Cortex-M4 cores in audio processing from the perspective of the instruction set. To this end, we first discuss some module code examples of different audio components, and then analyze the unique advantages of these processors in terms of loops and instructions. Finally, we will show the typical performance indicators of popular audio codecs and audio post-processing components to support these. The audio capabilities of the processor core.

Function block of audio processing module

In order to analyze the requirements of the audio processor, let us start with the functional blocks involved in the audio processing module, namely the audio codec and the audio post-processing component. The modules in the block diagram of this section are color-coded into 3 different categories. Green indicates a multiply-add (MAC) dense module, red indicates a mixed MAC and control code module, and blue indicates a control code module.

Audio codec

The Internet provides a large amount of compressed audio data, which inevitably requires music players to support a variety of popular audio decoders. The processing function of the audio encoder involves a series of functional blocks, which we will review next.

Figure 1: Block diagram of a typical audio encoder. (Electronic system design)

Figure 1: Block diagram of a typical audio encoder.

Audio encoder: The purpose of the audio codec is to achieve audio data compression while striving to obtain fidelity sound quality under the constraints of a given bit rate. A typical encoder is shown in Figure 1. Generally speaking, the audio encoder uses the following three steps to complete the compression: First, adjust the audio data through data filtering. Then, the audio data is framed by windowing and overlapping, waiting for further processing; then, this data is transformed into the frequency domain to eliminate redundancy in the time domain. Using psychoacoustic principles, calculate the amount of quantization noise that may be introduced and inaudible; finally, the data is quantified and further lossless compression is achieved through entropy coding. This encoded data is tightly packed into a bit stream. Although the compression achieved is somewhat lossy, it is difficult for the hearing to distinguish nuances.

Audio decoder: Reverse the above process of the audio encoder, that is, the principle that the audio decoder restores audio data from the bit stream. The MP3 decoder is taken as an example for discussion here, and the high-level module decomposition is shown in Figure 2. The block diagram takes compressed audio in MP3 file format as input and outputs uncompressed audio in PCM format.

These modules are named after functions and represent the operations they perform

Figure 2: Block diagram of the MP3 decoder. (Electronic system design)

Figure 2: Block diagram of the MP3 decoder.

"Bit Stream Demux" module parses MP3 bit stream

"Entropy & Inv Q" module performs Huffman decoding and inverse quantization

The "IMDCT" module reversely modifies this data by discrete cosine transform

"Overlap and Add" module performs windowing, overlap and addition operations

"Synthesis Filter Bank" reconstructs time domain samples from filter bank domain data

Yuhai company develop and produce of various rings sizes, bearing variety of electrode and metallisation configurations. This variety of Rings is fabricated from various material formulations for applications such as high power, sensitivity, stability.


Features

·         •   electrode type on request

·         •   Surface roughness on request,  for specific demanding application

·         •   Thickness/Radial frequency tuning available on request

    •   Wide choice of PZT formulations



Applications include

•    Sensors

•    Welding

•    Ultrasonic cleaning

•    Precise Measurement

•    Ultrasonic scalpel

•     Cauterisation

•     Phacoemulsification

•    Therapeutic ultrasound

•    Accelerometers


Dimension range

  

         Outer diameter

         3-180mm

        Inner diameter

         1-150mm

         Thickness

         0.2-25mm






PZT Piezoelectric Rings

Piezo Ring,Pzt Ring,Piezoelectric Ceramic Ring,Piezo Ceramic Ring

Zibo Yuhai Electronic Ceramic Co., Ltd. , https://www.yhpiezo.com