[Smart Hardware] Offline Voice Bluetooth Speaker

      Who could resist a Bluetooth speaker that can perform voice recognition without relying on the internet? The motivation for this project stemmed from my unreliable home Wi-Fi, which had left my Tmall Genie gathering dust. I happened to learn about the Qiying Tailun CI1302 third-generation voice chip, which boasts an extremely high recognition rate. The voice wake-up phrase and response can be customized, and it even offers an offline voice Bluetooth speaker solution – perfectly meeting my needs! Let's take a look at this solution together; by learning it, you too can design your own Bluetooth speaker!
I. Circuit Analysis
      When designing the circuit, it's essential to understand the core chip. Here, we've chosen the Qiying Tailun CI1302 chip, with a maximum CPU frequency of 220MHz and an operating voltage range of 3.6V~5.5V. It features one IIS interface, three UART interfaces, one IIC interface, six PWM interfaces, and ten GPIO ports, allowing for independent control or integration with a microcontroller or other chips. This chip supports 300 offline command words and also features a command word self-learning function, allowing for easy modification of wake-up words and response phrases – a very convenient feature.
Qiying Tailun's official website happens to have an evaluation board with Bluetooth functionality. Downloading the evaluation board documentation from the website and referencing its schematic, I designed the following circuit:
      In the PCB design, ensure C10, C11, C12, and C15 are placed close to the chip. Serial port 0 is reserved for program download, and serial port 1 is used for Bluetooth communication. Digital and analog grounds are separated and connected using resistor R5 for a common ground. The power supply is a convenient everyday Type-C power supply, outputting 3.6V through an HT7336 step-down chip to power the Bluetooth chip. The microphone input is filtered by two 100nF capacitors.

      The Bluetooth chip used is the AB5607E chip from Zhongke Lanxun. The peripheral circuit is simple; a separate antenna circuit can be designed and directly connected to the voice recognition chip. For the 26MHz crystal oscillator, a 20pF ±20pPM oscillator should be selected to avoid malfunction.
      For good audio sound, an amplifier is essential. When designing, you can choose between a mono or stereo amplifier based on your needs. Here, we chose a commonly used and inexpensive Class D amplifier chip, the 8002A. R10 is a pull-up resistor. The audio signal passes through an RC filter and enters pin 4 of the chip. R15 is a feedback resistor connected to the output pin VO1, forming an inverting amplifier circuit. The amplification factor can be adjusted by changing the value of R15 to change the output volume. Here, you can adjust the sound amplification factor according to the actual selected speakers. The amplification factor is: -HPOUT*R13/R15+(-DACL-BL*R14/R15).

      The speaker chosen here is a 4Ω 3W speaker, which combines high, mid, and low frequencies and has decent sound quality. The key point is that this speaker is very cheap!


II. Shell Design
      After designing the PCB, you can directly design the shell file in JLCPCB EDA Professional Edition. Shell design becomes very simple here. You only need to combine the board frame to set its shell shape and size, make holes in the corresponding positions, and set screw posts for support. The designed model is shown in the image below:
 
III.
      3D Printing Which 3D printing company is the best? Check out JLCPCB's 3D Monkey! 3D Monkey 3D printing not only supports resin and nylon, but also metal, color printing, and the well-known FDM printing. There's always a process that meets your needs. It's worth mentioning that the 3D Monkey platform currently offers monthly coupons. For details, please click the following link:
Click to view 3D Monkey free printing.

IV. References
      If you want to learn more about offline speech recognition technology, you can directly visit the Qiying Tailun official website to explore. It contains a large number of application solutions and firmware for learning and reference:
https://aiplatform.chipintelli.com/