WTK6900FC-24SS Voice Recognition Chip
AI‑Powered Speech Recognition with 5‑8m Far‑Field, 300 Offline Commands & 95dB SNR Audio
The WTK6900FC-24SS is a specialized AI voice recognition chip designed for high‑performance speech processing. Powered by a 220MHz CPU with a dedicated BNPU V3 neural network accelerator, it supports DNN-HMM deep learning algorithms for 5–8 meter far‑field voice recognition with up to 300 offline commands. With 95dB SNR audio codec, built‑in 2MB Flash and 640KB SRAM, and UART control, it is the ideal solution for smart home, industrial, and automotive voice control applications.
- 220MHz AI core – high‑performance processing
- 5–8m far‑field – robust voice pickup
- 300 offline commands – no cloud required
- 95dB SNR – high‑quality audio input/output
- DNN-HMM – deep learning voice recognition
WTK6900FC-24SS Key Features
AI & Recognition Specifications
| Parameter | Value / Description |
|---|---|
| Processor Core | 220MHz CPU + BNPU V3 Neural Network Accelerator |
| AI Algorithms | DNN, TDNN, RNN, CNN, Parallel Vector Operations |
| Recognition Technology | DNN-HMM Deep Neural Network |
| Far‑Field Range | 5 – 8 meters |
| Offline Commands | Up to 300 command words |
| Languages Supported | Chinese, English, Japanese, and more (customizable) |
| Voice Detection | Local voice activity detection (VAD) & wake‑word |
| Noise Reduction | Deep learning‑based noise suppression |
| Command Learning | Supports self‑learning of command words |
Audio Codec Specifications
| Parameter | Value / Condition |
|---|---|
| ADC Resolution | 16‑bit |
| ADC SNR | ≥ 95 dB |
| DAC Resolution | 16‑bit |
| DAC SNR | ≥ 95 dB |
| Audio Interfaces | 1×IIS (Master/Slave), 1×PDM (2‑channel) |
| Microphone Input | ECM (Electret Condenser Microphone) |
| Recommended MIC | JMO-627BA283R-10TC (Sensitivity: -28dB, S/N: 70dB) |
| ADC Channels | Differential input (MICPR / MICPL) |
| DAC Output | Single‑ended (DAC pin) |
Technical Specifications
| Parameter | Value / Condition |
|---|---|
| Package | SSOP24 |
| CPU Frequency | Up to 220 MHz |
| Flash Memory | 2MB (WTK6900FC) / 32MB (WTK6900FC3) |
| SRAM | 640 KB |
| eFuse | 512‑bit for encryption |
| PMU Input Voltage | 3.6V – 5.5V |
| VOUT (IO / Codec) | 3.3V (2.97–3.63V) |
| VREG (Core) | 1.1V (0.99–1.22V) |
| GPIO | 10 high‑speed (20MHz), 7×5V‑tolerant |
| UART | 3×, up to 3Mbps, 5V‑tolerant |
| I2C | 1× |
| Timers | 4×32‑bit timers, 2×watchdog |
| PWM Channels | 6× |
| ADC | 1×12‑bit SAR ADC |
| Operating Temperature | ‑40°C ~ +85°C |
| Storage Temperature | ‑65°C ~ +150°C |
Key Pin Functions (SSOP24)
| Pin | Name | Function |
|---|---|---|
| 1 | AVDD | Analog power (4.7µF cap to ground) |
| 2 | VCC | Power input (3.6–5.5V) |
| 3 | VOUT | 3.3V LDO output (4.7µF cap) |
| 4 | VREG | Core voltage (4.7µF cap) |
| 9 | P02 | UART1_RX (communication input) |
| 10 | P03 | UART1_TX (communication output) |
| 14 | TX / DAT | UART1_RX / Programming |
| 15 | RX / CLK | UART1_TX / Programming |
| 16 | BUSY | Busy signal output |
| 17 | DAC | DAC audio output |
| 20 | MICPR | Microphone input (−) |
| 21 | MICPL | Microphone input (+) |
| 22 | MICLDO | Microphone power supply |
| 23 | VCM | Common mode voltage reference |
| 24 | AGND | Analog ground |
Application Scenarios
Why Choose WTK6900FC-24SS
Frequently Asked Questions
What is the WTK6900FC-24SS?
WTK6900FC-24SS is an AI voice recognition chip with a 220MHz CPU, BNPU V3 neural network accelerator, built-in 2MB Flash, 640KB SRAM, 95dB SNR audio codec, and support for 5–8 meter far-field voice recognition with up to 300 offline commands.
What is the recognition range?
WTK6900FC-24SS supports 5–8 meter far-field voice recognition, making it suitable for smart home, industrial, and automotive applications.
How many offline commands are supported?
WTK6900FC-24SS supports up to 300 offline command words, enabling voice control without cloud connectivity.
What languages are supported?
WTK6900FC-24SS supports multiple languages including Chinese, English, Japanese, and more, with flexible customization options.
What AI algorithms are used?
The WTK6900FC-24SS chip uses DNN-HMM deep neural network-based speech recognition, powered by the BNPU V3 neural network accelerator, supporting DNN, TDNN, RNN, CNN, and parallel vector operations.
What control interface does it use?
WTK6900FC-24SS uses a standard UART interface (3.3V TTL) with commands for version query and voice command output. The chip outputs recognized command IDs via UART.
What is the operating voltage range?
The PMU input voltage ranges from 3.6V to 5.5V, with VOUT at 3.3V for IO and codec power.
Does WTK6900FC-24SS have built-in Flash memory?
Yes. The WTK6900FC-24SS chip includes 2MB of built-in Flash memory and 640KB of SRAM. A 32MB Flash version (WTK6900FC3) is also available.
What microphone does WTK6900FC-24SS support?
WTK6900FC-24SS supports electret microphones (ECM) with a sensitivity of -31 to -25dB and an S/N ratio of 70dB. Recommended model: JMO-627BA283R-10TC.
What are the typical applications?
Smart home appliances, smart lighting, smart bathrooms, smart furniture, smart toys, industrial controls, automotive voice systems, and any application requiring local voice control and recognition.
Important: The chip supports both internal RC oscillator and external crystal oscillator. For microphone input, use a differential connection (MICPR/MICPL) with an electret microphone. The BUSY pin indicates recognition/processing status. VCC input range is 3.6–5.5V – ensure stable power supply. The DAC output provides analog audio output for TTS playback or voice prompts. For custom command sets, contact Waytronic for firmware customization.
