MCU Speech Processing Technology

Application of SCM Speech Processing Technology

Multimedia technology is an extremely active emerging technology in modern computer application technology. Audio-visual processing is the core research object of multimedia technology, and speech processing is an extremely important practical technology in sound processing technology. The basis of speech processing technology is phonetics and digital signal processing, which includes speech analysis technology, speech storage technology, speech recognition technology and speech synthesis (reproduction) technology. This article takes Sunplus SPCE061A as an example to discuss the principle and application of single-chip voice processing technology.

1. Audio processing

Computer processing audio mainly adopts several technologies: sampling and quantization technology, audio preservation technology, audio compression coding technology, audio reproduction and identification technology. Sampling and quantization is to collect the sounds in nature into the computer according to certain requirements, which is the basis of sound processing technology. The frequency of audible audio is in the range of 20Hz~20kHz, and the frequency of voice is generally in the range of 60Hz~500Hz.

Audio can be divided into three types: waveform sound, speech and music. Like any sound, speech is also manifested as a waveform sound, but the waveform sound cannot express the connotation of language and speech. Speech is an abstraction of speech sound, the carrier of language, and an information system unique to human society. Music is a special sound, a weak subset of sound, and a normalized and symbolized sound. But music cannot symbolize all sounds. A musical score is a group of symbols that symbolize sound, and different combinations represent more complex sound information than a single number.

When a computer processes audio, it must first digitize (discrete) the analog (continuous) sound waveform, which is achieved through an A/D converter. The converted audio is called digital audio, and its quality depends on the A/D conversion The conversion rate and resolution of the converter, the higher the rate, the more data collected per unit time, the higher the resolution, the higher the accuracy of the collected data, so the audio sampling frequency and quantization digits in audio processing are two an important parameter. The collected audio signal is saved by audio file. Audio files are divided into two categories: sound files and MIDI files.

2. SCM voice processing technology

In terms of single-chip voice processing, the usual solution adopts two approaches: one is to expand the design of the single-chip itself, and place the voice processing function module directly in the single-chip; the other is to expand the functions of the single-chip with the help of a special voice processing chip. Due to the large number of special voice processing chip series and single function, it is necessary to add a lot of peripheral devices to realize a complete system, so it is not very convenient to use. Therefore, comparing these two solutions, the first one has obvious advantages, but as For general-purpose processors, this solution is inappropriate.

Chips specially designed for speech recognition and speech processing applications need a high-speed DSP core processor to undertake control and computing tasks, and also need an appropriate amount of on-chip integrated ROM/RAM memory to store applications, drivers, various data and Speech recognition, speech compression algorithms and constant tables used in the algorithms, etc., which increase the cost of chip design and manufacturing. Because the integrated ROM/RAM memory of the single-chip voice processing chip cannot be very large, the single-chip microcomputer is not suitable for large-scale voice processing systems.

3. Speech processing technology of Sunplus SPCE061A MCU

3.1 Sunplus SPCE061A MCU

Sunplus SPCE061A is a microcontroller with 16-bit μ’nSP structure. The chip has a hardware multiplier, which can realize complex operations such as multiplication and inner product operations. Sunplus SPCE061A single-chip microcomputer is advanced in design, outstanding in features, easy to learn and use, and reflects the new trend of modern microcontroller industry development.

3.2 Realization of Speech Processing by Sunplus SPCE061A MCU

Sunplus SPCE061A makes functions related to speech processing into function modules, which are realized through API calls. These functions are divided into two categories: Sunplus Audio Coding (SACM) for audio data playback and Speech Recognition (BSR) for recognizing speech.

After the WAVE files generated by input are compressed into corresponding voice data such as SACM_A2000 or SACM_S480 according to different compression algorithms, these data can be processed by SACM-LIB library. The library makes A/D, encoding, decoding, storage and D/A into corresponding modules. Each module has its application program interface API function, and the respective functions can be realized by calling the corresponding function.

For example, for speech recognition of a specific person, the SACM-LIB library provides API functions defined in two formats including C language and assembly language. The commonly used speech recognition API functions are:

(1) SRAM initialization function int BSR-DeleteSD-Group (int); (2) training function int BSR-Train (int CommandID, int TrainMode); (3) recognition initialization function int BSR-InitRecognizer (int AudioSource); (4 ) Get the recognition result function int BSR-GetResult (void); (5) Stop the recognition function void BSR-StopRecognizer (void); (6) Identify the interrupt program-BSR-FIQ-Routine;

With the relevant API functions of voice compression and playback, a complete voice application system can be formed.

Summarize

Voice processing technology has an important impact on the design technology of the next generation of multi-mode interactive human-machine interface. With the rapid increase in the demand for high-performance and high-stability voice interfaces in consumer electronics products, single-chip microcomputers and embedded voice processing technologies are developing rapidly. . However, there are still some problems in the solution of voice processing technology. With the further research of software and hardware technology, these problems will be reasonably solved in the near future. Voice processing technology will bring a wider space for the in-depth application of computers. .

The above is the application of single-chip voice processing technology introduced by Shenzhen Zuchuang Microelectronics Co., Ltd. for you. We have rich experience in customized development of smart electronic products, can evaluate the development cycle and IC price as soon as possible, and can also calculate the PCBA quotation. We are the agent of Sonix MCU and Yingguang MCU agent, selling and developing MCU and voice IC solutions of Sonix and Yingguang. We act as an agent and develop ICs and solutions for Jieli, Ankai, Allwinner, Realtek, etc., and also develop BLE Bluetooth ICs, dual-mode Bluetooth modules, wifi modules, and IoT modules. We have hardware design and software development capabilities. Covering circuit design, PCB design, single-chip microcomputer development, software custom development, APP custom development, WeChat official account development, voice recognition technology, Bluetooth development, wifi technology, etc. It can also undertake the research and development of smart electronic products, the design of household appliances, the development of beauty equipment, the development of Internet of Things applications, the design of smart home solutions, the development of TWS earphones, the development of Bluetooth earphone speakers, the development of children's toys, and the research and development of electronic education products.