The XMOS xcore-voice solution is a complete product built on the xcore platform's hardware, software, and tools.¶
This solution offers a sample design of a voice pipeline using XMOS's industry-proven audio front end, including far-field voice processing and support for third-party ISV voice algorithms such as automatic speech recognition (ASR) engines for keyword detection or local command dictionaries.
This enables various applications and end products, such as smart TVs, set-top boxes, and smart home appliances. Specifically, the xcore-voice solution allows product designers to quickly and cost-effectively implement "across-the-room" voice interfaces while achieving optimal audio quality.
The XMOS xcore-voice solution is a complete product built on the xcore platform's hardware, software, and tools.
This solution offers a sample design of a voice pipeline using XMOS's industry-proven audio front end, including far-field voice processing and support for third-party ISV voice algorithms such as automatic speech recognition (ASR) engines for keyword detection or local command dictionaries.
This enables various applications and end products, such as smart TVs, set-top boxes, and smart home appliances. Specifically, the xcore-voice solution allows product designers to quickly and cost-effectively implement "across-the-room" voice interfaces while achieving optimal audio quality.
XCORE®-VOICE Solution Based on XCORE.AI - Next-Generation Intelligent Solutions for Smart Voice Applications
The xcore.ai is the third generation of the xcore processor architecture, providing unprecedented flexibility by combining general-purpose DSP (fixed and floating point), AI (32b, 8b, and 1b networks), and IO processing (nanosecond resolution). Unlike traditional SoCs, xcore systems can be architected in software, avoiding the typically lengthy hardware development cycles associated with custom silicon chips.
The sample designs in the xcore-voice solution leverage the processor's capabilities through xcore platform software, providing building blocks for developing various applications. The xcore platform software supports development in the C programming language—on bare metal or within FreeRTOS.
Sample designs are available in source code or precompiled form and can be quickly evaluated on the XK-VOICE-L71 voice evaluation kit.
xcore.ai and the voice evaluation kit are available through XMOS sales and XMOS general distribution partners.
Key Highlights¶
Voice Processing Components¶
Two PDM microphone interfaces
Digital signal processing pipeline
Full-duplex stereo acoustic echo cancellation (AEC)
Reference audio via I2S with automatic bulk delay insertion
Point noise suppression through an interference canceller
Switchable static noise suppressor
Programmable automatic gain control (AGC)
Flexible audio output routing and filtering
Independent audio paths for communication and automatic speech recognition (ASR)
Support for Wanson speech recognition or user-defined third-party ASR
Device Interface Components¶
Full-speed USB2.0 device supporting USB Audio Class (UAC) 2.0
Flexible peripheral interfaces
Programmable digital general-purpose inputs and outputs
Sample Designs Utilizing Components¶
Far-field voice local commands (FFD)
Far-field voice assistants (FFVA)
Firmware Management¶
Boot from QSPI flash
Default firmware image for power-on operation
Option to boot from a local host processor via SPI
Device firmware updates (DFU) via USB or other transport methods
Power Consumption¶
Typical power consumption 300-350mW
As low as 55mW in low power mode (using DEMO VNR)
Evaluation/Development Kits and Boards¶
Support¶
for XCORE® platform software
XTC tools
Software-defined hardware platform
Support and training materials to accelerate time to market and reduce risk Turnkey sample designs and support peripherals
Voice application boards and support peripherals
Application-specific software components such as tunable voice algorithms and third-party voice models (licensable)
Core application software components such as IO libraries, configuration libraries, math libraries, optional FreeRTOS kernel
Compiler toolchain, assembler
Bootloader, debugger, simulator, board support
xcore.ai product series based on the third-generation XMOS architecture XS3
Evaluation Kit¶
The XK-VOICE-L71 voice evaluation kit can be used with the Raspberry Pi HAT for integration with sample AVS clients or as a standalone USB accessory for host systems.
Features¶
XU316-1024-QF60A-C24 xcore.ai processor
Raspberry Pi HAT connector
2 Infineon IM69D130 MEMS microphones
71mm microphone spacing
Microphone mute switch
Speaker output (line level)
USB / I2S host interface support
Part Number Descriptions¶
XU316-1024-QF60A-C24: 1.8V IO XCORE.AI processor
XU316-1024-QF60B-C24: 3.3V IO XCORE.AI processor
XK-VOICE-L71: Voice evaluation kit
XMOS China Contact: Wilson Xiao(Application Solution and Business Cooperation - Senior Manager)
Email: Wilsonxiao@Xmos.com
Phaten XMOS Solution Contact:
Allen Su
Email: hua@phaten.com