Speech recognition chip, speech control chip, nrk10 recognition chip
summary
Nrk10 speech recognition chip is a high-performance and low-cost offline speech recognition chip independently developed. It has the functions of speech recognition and broadcasting. It needs to plug-in SPI flash to store entries or speech broadcasting content. It has the advantages of high recognition rate, industrial performance, simple and easy to use, convenient to update entries and so on. It is widely used in smart home, AI, artificial intelligence, toys and other fields.
2. Product features:
• working voltage: 3.3V ~ 5.5V, generally 4.2V
• sleep current: < 7ua
• non dormant standby current: 25 ~ 50mA
• wake up mode: after sleep, you can wake up from any GPIO, RTC, or WDT interrupt to make it enter the standby state; In standby mode, wake up words or entries to enter the working state.
• fixed entry, non-specific person identification
• it can recognize 20 entries (three words or more are recommended for each entry), and the total number of words is between 50 ~ 60 words.
• identification environment: quiet without echo
• recognition effect: in quiet echo free environment, the recognition rate can reach 90% or more within 2 meters, and the maximum distance can reach 5 meters
• language recognition: 32 languages can be recognized, such as English, Chinese, Japanese, Cantonese, etc. (multiple languages cannot be recognized at the same time, that is, one module cannot recognize Chinese, English and Japanese at the same time)
• PWM output, which can directly drive 8 Ω 0.5W horn; DAC output, external power amplifier.
• the audio output content needs to be placed in the external SPI_ In flash
• low voltage reset: reset the chip below 1.6V
• built in low dropout voltage regulator (LDO) - can provide 25mA load current at 3.3V - configurable output voltage: 8 options: 1.5v/1.7v/1.8v/2.4v/2.5v/2.7v/3v/3.3v, the default is 3.3V.
• operating temperature: - 10 ° C ~ + 70 ° C
• the chip program and entry storage mode is OTP storage mode. • one time burning is stored in the main control chip to facilitate sample and mass production. The broadcast voice is stored in external SPI flash. Support 8m / 16m / 32mbit SPI flash.
• built in UART serial port to facilitate MCU control and communication
• built in mic 2-stage bias amplifier circuit, self setting AGC gain control and rate noise circuit.
• packaging forms include lqfp32 and cob
3. Pin
3.1 chip pin diagram
Nrk10 speech recognition chip, offline speech recognition chip, person specific speech recognition chip
(nrk10 chip lqfp32 pin diagram without serial port version)
3.2 chip pin description table (nrk10 chip lqfp32 pin diagram without serial port version)
Pin name function description
1pb9io port for input and output
2pb8io port for input and output
3pb7 / xoutio port / crystal oscillator output pin, crystal oscillator frequency range 32768Hz ~ 16mhz
4pb6 / xinio port / crystal oscillator output pin, crystal oscillator frequency range 32768Hz ~ 16mhz
5PB5/SPIO_ Io3flash communication pin connected to IO3 / hold
6PB4/SPI0_ Io2flash communication pin connected to WP / io2
7PB3/SPIO_ Io1flash communication pin connected to do / IO1
8PB2/SPI0_ Io0flash communication pin connected to di / IO0
9PB1/SPIO_ Sckflash communication pin connected to SCK
10PB0/SPIO_ Csbflash communication pin connected to CS
11SPI0_ Vddflash power supply VCC
12vdd-adc analog power string resistance to VCC
13vmic microphone power supply pin
14micp microphone P end (near the power end)
15micn microphone N end (ground end)
16MICOUTMICOUT
17OPIOPI
18AGCAGC
19vss1 ground
20rstb / VPP programming power supply, leave blank when using
21pa3 / ain3 / sckio port / programming data pin, leave blank when using
22pa4 / ain4 / TMO / sdoio port / programming data pin, leave blank when using
23pa5 / ain5 / IR / sdiio port / programming data pin, leave blank when using
24vdd1 power supply
25pa10 / pwma2io port, pulse width modulation output signal pin
26pa11 / pwma3io port, pulse width modulation output signal pin
27pa12io port can be used as the communication line of MCU
28pp1 / dac0pwm1 / DAC signal output pin, horn output port
29vdd2 power supply
30pp2pwm2 signal output pin, horn output port
31vss2 ground
32pb10io port for input and output
4. Function introduction
4.1 introduction to UART serial port function
Nrk10 speech recognition chip has built-in standard UART asynchronous serial port interface, and the communication data format is: start bit: 2 bits; FA； Data bit: 2 bits; 00H~FFH； Check bit: 2 bits. The sum of check bit and data bit is FFH. To use the computer serial port debugging assistant, you need to correctly set the parameters of the serial port, as shown in the figure:
Nrk10 speech recognition chip, offline speech recognition chip, person specific speech recognition chip
4.2 introduction to playback control
Entry introduction: the entry refers to the entry in the standard version of the recognition instruction that we want to say, "Xiao Ming" is used as the wake-up word, and other entries can be recognized only after wake-up. See the following table for details:
Nrk10 speech recognition chip, offline speech recognition chip, person specific speech recognition chip
Note: 1. This version can be directly connected to the lamp without serial port chip PA10 and PA11; PA6 and PA7 with serial port can also be used to directly output serial port signals. 2. The Standard Version entries are used for testing. If you want to modify the entries, please contact the business personnel. 3. UART port can be used to communicate with MCU. 4. The delay time after wake-up is 10 seconds. After receiving instructions, it is necessary to re time without calling the wake-up word every time.
5. Circuit diagram
Nrk10 speech recognition chip, offline speech recognition chip, person specific speech recognition chip
(circuit schematic diagram without serial port version)
Nrk10 speech recognition chip, offline speech recognition chip, person specific speech recognition chip
(with circuit schematic diagram of serial port version)