“The outbreak of the global epidemic has accelerated the development process of digital transformation and intelligent Internet of Things. In order to effectively fight the epidemic and reduce people’s direct contact in daily life, non-contact technology is widely used in various scenarios. Among them, speech recognition technology, as a non-contact technology, interacts with devices through voice or voice commands, and has attracted much attention in the post-epidemic era.
“
On March 16, 2022, Dalian General Holdings, a leading semiconductor component distributor dedicated to the Asia-Pacific market, announced that its subsidiary Pinjia has launched a Wi-Fi 6 AIoT edge computing voice based on MediaTek’s Filogic 130A (MT7933). Identify the scheme.
Figure 1 – Display board diagram of Dalian Dapinjia’s Wi-Fi 6 AIoT edge computing voice recognition solution based on MediaTek products
The outbreak of the global epidemic has accelerated the development process of digital transformation and intelligent Internet of Things. In order to effectively fight the epidemic and reduce people’s direct contact in daily life, non-contact technology is widely used in various scenarios. Among them, speech recognition technology, as a non-contact technology, interacts with devices through voice or voice commands, and has attracted much attention in the post-epidemic era. Based on this background, Dalian Dapinjia has launched a Wi-Fi 6 AIoT edge computing voice recognition solution based on MediaTek Filogic 130A (MT7933). The solution combines advanced Wi-Fi and Bluetooth capabilities with the latest voice processing and power management technologies to provide new design ideas for smart speakers, smart homes, home entertainment and automotive multimedia entertainment.
Figure 2 – Scenario application diagram of Dalian Dapinjia’s Wi-Fi 6 AIoT edge computing voice recognition solution based on MediaTek products
MediaTek’s new wireless networking system-on-chip Filogic 130A (MT7933) integrates a microcontroller, AI engine, Wi-Fi 6 and Bluetooth 5.2, power management unit (PMU), independent audio digital signal processor (DSP) and other units. Among them, the audio digital signal processor (DSP) enables device manufacturers to easily add voice assistants and other services to their products. With advanced functions and high level of integration, this solution can provide energy-saving, reliable and efficient network connectivity for small-sized devices, making it an excellent choice for various Internet of Things (IoT) devices.
Not only that, the Voice Activity Detection (VAD) technology in the Filic 130A is also very intelligent. When it detects human speech, it will automatically ignore the silent segment in the audio, and will only perform audio processing after hearing the human speech to achieve the purpose of low power consumption. Whether it is a single microphone with a simplified design or multiple matrix microphones, the Filic 130A can perform AEC (Acoustic Echo Cancelling), far-field processing (Far-Field Process) and other functions to enhance speech recognition.
Figure 3 – Block diagram of Dalian Dapinjia’s Wi-Fi 6 AIoT edge computing voice recognition solution based on MediaTek products
In addition, the Filic 130A also supports native voice commands. With pre-defined voice commands, even in the absence of network connection and network delay, the device can be easily controlled by voice commands. Such as: control lights, volume, and audio controls such as play, pause music, and pre and post.
Core technical advantages:
• MediaTek’s new wireless networking system-on-chip Filogic 130A (MT7933), which integrates an independent audio digital signal processor (DSP), can easily add services such as voice assistants to products. The HiFi4 DSP used has 3 ADC/2 DACs and dedicated SRAM to provide ultra-low power, Always-On microphone functionality with Voice Activity Detection (VAD) and Wake Word support.
• Dalian Dapinjia Group provides all-round technical support, from early stage development and design of suitable audio hardware, addition or deletion of pre-defined native voice commands and adjustment of audio processing performance. Wireless connection performance test in mass production, etc.
Program Specifications:
• Application processor:
ARM®Cortex-M33 MCU with floating point operation, operating clock 300MHz;
1MB embedded SRAM and 8MB virtual SRAM (PSRAM);
Support external serial flash up to 16MB, support in-place execution (XIP);
Network security hardware encryption engine includes AES, DES/3DES, SHA, ECC, TRNG;
Support 47 groups of GPIOs multiplex switch SPI, I2C, Aux ADC, UART, and GPIO functions;
12 DMA channels are supported.
• Audio Digital Signal Processor (DSP):
Cadence®Tensilica®HiFi4 processor, operating clock 600MHz;
Audio Codec has 2 sets of ADC and 1 set of DAC;
256KB embedded SRAM memory;
Voice Activity Detection (VAD) and Wake Word;
3.5mm audio port for external active speakers.
• Wi-Fi technical specifications:
Dual-band IEEE 802.11 1T1R a/b/g/n/ax 5GHz and 2.4GHz;
2.4G/5GHz frequency band, 20MHz bandwidth MCS0 ~ MCS8.
• Bluetooth technical specifications:
Compliant with Bluetooth v5.0, transmission rate 2Mbps PHY, support long-range Long-range and LE Advertising Extensions.
The Links: DMF5005NY-LY BSM100GT120DN2