Classification and basic principle of speech recognition chip

Speech recognition chip is also called speech recognition IC. Compared with traditional speech chip, the biggest feature of speech recognition chip is speech recognition. It can make the machine understand human speech, and can perform various actions according to commands, such as blinking eyes and moving mouth (intelligent baby). In addition, the speech recognition chip also has the function of recording and playback with high quality and high compression rate, which can realize man-machine dialogue.
The technologies involved in speech recognition chip include: signal processing, pattern recognition, probability theory and information theory, phonation mechanism and auditory mechanism, artificial intelligence and so on.
Classification of speech recognition chips
According to the restrictions of users, speech recognition chips can be divided into person specific speech recognition chips and non person specific speech recognition chips.
The person specific speech recognition chip is for the speech recognition of the specified person. If other people's words are not recognized, the user's speech reference samples must be stored in the database as comparison, that is, the person specific speech recognition must be speech trained before use. Generally, it can be used after training the speech entries twice according to the machine prompt.
Speaker independent speech recognition does not need the recognition technology for the specified person, regardless of age and gender, as long as you speak the same language. The application mode is to collect the voice samples of about 200 people according to the determined more than a dozen voice interactive entries before the product is finalized, process the PC algorithm to obtain the voice model and feature database of the interactive entries, and then burn them on the chip. Machines using this chip (smart dolls, electronic pets, children's computers) have interactive functions.
Some speaker independent speech recognition applications are phoneme based algorithms. In this mode, interactive recognition can be done without collecting many people's voice samples, but the disadvantage is that the recognition rate is not high and the recognition performance is unstable.
Basic principle of speech recognition chip
All embedded speech recognition systems adopt the principle of pattern matching. The input speech signal is first preprocessed, including speech signal sampling, anti aliasing filtering and speech enhancement. Next, feature extraction is used to extract one or more groups of parameters that can describe the characteristics of speech signal from the speech signal waveform.
The data after feature extraction is generally divided into two steps. The first step is the system "learning" or "training" stage. The task of this stage is to build a reference pattern library. Each word in the thesaurus corresponds to a reference pattern. It is obtained by repeating the word for many times, and then through feature extraction and some training. The second is the "recognition" or "test" stage. According to certain criteria, the speech feature parameters to be tested and the distortion measure between the speech information and the corresponding template in the pattern library are obtained. The most matching is the recognition result.

Commax-Tech Electronic Co., Ltd      Electronic component specialist

B23, second floor, ASEAN building, Longhua District, Shenzhen

sales@commax-tech.com

https://commax-tech.com

Keyword：Speech Recognition Chip   Signal Processing   Pattern Recognition   Probability Theory   Information Theory   Sound Mechanism   Hearing Mechanism   Artificial Intelligence   Smart Doll   Electronic Pet   Children's Computer   Commax-Tech Electronic