RSC-4X series speech recognition integrated application circuit design

MOS power IC full range

MP24AE off-the-shelf supply

RSC-4x series is a new generation of embedded speech recognition processor integrated circuit, which has the characteristics of high integration, simple peripheral circuit, low power consumption, no fear of power failure and convenient use. A complete speech recognition system consists of only the RSC-4x series of chips and a small number of external components (speakers, microphones, audio input/output circuits, memory and battery or rectified regulated power supplies). RS C-4 x series chips for 7.0 speech recognition technology RSC-4x series chips are usually packaged in 48P/64P/100P-LQFP.

How it works: The RSC-4x is an interactive intelligent speech recognition circuit that processes speech recognition by executing embedded Markov chains and neural networks. The speaker needs to store the speech recognition information in the chip's external memory (such as sRAM, EEPROM, flash memory, etc.), and the ROM is required to store the word to be recognized. The RSC-4x has several unique recognition features:

1. Speech recognition does not require training. 16 words can be identified in one active unit (the words stored in the unit are limited by the internal ROM and external memory).

2. Receive each special word continuously. Thanks to this function, this product can be used in general situations and in quiet environments for some special words.

3. Word fixed point measurement allows the chip to receive up to 5 SI or 10 SD at a time. In this word fixed mode, there is no need to maintain a quiet environment. The RSC-4x family can store up to six sD words in the chip's sRAM or more words in external memory.

The RSC-4x provides high-quality speech synthesis using clever frequency domain rules. Typical data rates are less than 5k/s. Speech synthesis requires on-chip or off-chip ROM to store the speech data to be synthesized. The RSC-4x offers four high quality music/speech synthesis methods. The RSC-4x is also unique in that it can use a computer network system to produce high-end music with 128 sounds and 47 real instrument sounds. The MIDIRSC-4x can record and replay sound at different sampling ratios. The sampling ratio depends on the voice. The duration and sound quality requirements, when the data rate is 14k / s, generally get a higher sound quality. The RSC-4x also eliminates static differences to improve sound quality and reduce storage capacity.