Manually controlled speech recognition system circuit diagram

Single chip microcomputer STM32L151CCU6

Crystal oscillator

Industrial Router Crystal 3.2*2.5mm 3225 26M (26.000MHZ) 12PF 10PPM 20PPM 30PPM

There are many speech recognition chips. The following describes the speech recognition integrated circuit with 756410 as an example. The TSC410 speech recognition chip recognizes 40 fields and produces 40 different control signals (in binary form) that can be used for most voice control needs. This chip is widely used in the automatic control of sound, multimedia technology applications and computer voice command input.
The 756410 voice recognition chip is composed of CPU, A/D conversion, ROM, voice amplifier circuit, compression circuit, filter circuit, oscillator and interface interface. The TSC410 voice recognition chip is characterized by a single-chip structure, few peripheral components, large recognition capacity, strong confidentiality, short response time, diverse operation modes and flexible segmentation modes. The 756410 is generally operated in two modes, manual control and CPU control. The CPU control mode needs to be connected to the MCU and its expansion circuit. The space is not covered here. Please refer to other bibliographies. The following uses the manual control method as an example to introduce the application circuit of 756410 and its working principle. As shown in Figure 1, the schematic diagram of the 756410 manual control is shown.
The working process of the speech recognition circuit is usually divided into two steps: 1 to establish a sample; 2 to identify a sample. The so-called establishment of a sample refers to the process of inputting a specific voice. For example, if you can input the voice “Please enter” in the microphone, after the voice chip receives the “Please enter” voice, it will be amplified, shaped, and then bandpass filtered to filter out unnecessary high frequency and low frequency. The component, after being compressed, is sent to an A/D converter to turn it into a digital signal. This can ensure that the CPU can reduce the amount of data processed by the CPU under the premise of sufficient discrimination accuracy to save RAM space and improve the speed of chip-to-speech recognition. After the above process, under the control of the CPU, the voice signal of the input person is stored in the voice sample area in the form of a digital signal, that is, the first step is established.
Manually controlled speech recognition system circuit design

Figure 1 756410 Manually Controlled Circuit Schematic Note that when inputting a voice, people should not be too close to the microphone when speaking, nor should they be too far away from the microphone. The volume should not be too large when speaking, so it should not be too small to avoid distortion. It is not conducive to the establishment of samples. When the sample is created, if there is voice re-enter the microphone, the chip begins to identify the sample. If you say "please enter", the chip will process the speech (amplification, shaping, filtering, compression, A/D conversion), convert it into a digital signal, compare it with the created samples one by one, when you find the voice When the signal is the same as a sample signal, the chip outputs the area code of the area, that is, a data can be obtained from the data bus of the chip, and after the corresponding processing of the data, the corresponding actuator action can be driven, for example, The door will open automatically, please enter the room. When the voice signal of the input person is compared with the established sample one by one, and the sample signal identical to the voice signal is not found after the comparison, the chip sends a signal indicating that the failure has occurred. This is the second step to identify the sample.