Today, we the launch the first single System-on-Chip (SoC) development kits for the Alexa Voice Service (AVS). We are excited to broaden the portfolio of development kits for AVS to include complete single-chip solutions that provide a simplified architecture, high performance, and low total BOM cost. These new System Development Kits for AVS are ideal for OEMs and ODMs looking for a complete solution to build Alexa-enabled products.
Why Single-Chip SoC Solutions
A typical system designed for AVS requires a microphone and speaker with three processing blocks: an audio front end (AFE) to clean up the input speech, a wake word engine (WWE) to recognize the voice command for “Alexa,” and an AVS client to send utterances to and receive directives from the cloud.
Existing Audio Front End Development Kits for AVS utilize a dedicated digital signal processor (DSP) to implement the AFE processing block, while leveraging a Raspberry Pi to host the WWE and AVS client for prototyping purposes. This provides a modular solution to easily integrate voice capture abilities into connected products. OEMs can use a qualified solution to easily add Alexa voice capabilities, and choose the production-ready SoC they prefer for their final product.
System Development Kits for AVS offer a complete, end-to-end system reference design for building AVS products. This category includes AFE development kits that are also available in qualified bundles with production-ready SoCs. The new System Development Kits for AVS that we’re introducing today have SoCs that perform all three processing block functions on a single chip. They allow OEMs to leverage existing SoC relationships and existing infrastructure, while minimizing the impact on their design cycles. The single-chip SoC development kits in this category do not require an external DSP.
“By qualifying these first single SoC solutions, we’re expanding the portfolio of development kits for AVS in an important way for OEMs and ODMs building Alexa-enabled products,” said Priya Abani, Director, Amazon Alexa. “These dev kits reduce the cost of integrating Alexa, as well as accelerate time to market due to a near-production-ready architecture that streamlines technology integration and accelerates product development.”
Introducing single-chip SoC development kits from Qualcomm, Amlogic, and Allwinner
The first single-chip SoC solutions qualified by Amazon are development kits from Qualcomm, Amlogic, and Allwinner:
- Qualcomm's Smart Audio 6-Mic Development Kit for Amazon AVS is based on the Qualcomm® Smart Audio platform. It supports the development of Alexa-enabled table-top smart home devices and premium smart speakers. The kit includes Qualcomm’s Snapdragon™ application processor and system software; Fluence IoT technology for beamforming, echo cancellation, noise suppression, barge-in support, and far-field voice capture in high-noise environments; integrated audio decoding and post-processing algorithms to support premium sound quality; WWE tuned to “Alexa”; and AllPlay™ multi-room networking capabilities.
- Amlogic's A113X1 Far-Field Development Kit for Amazon AVS is designed for commercial developers integrating Alexa into smart speakers, smart home hubs, and other devices. The development kit comes with a board that is configurable for 2-mic linear, 4-mic triangular, and 6-mic circular arrays, making it a flexible solution for a range of hands-free and far-field applications. The SoC supports advanced algorithms for echo cancellation, beamforming, noise reduction, as well as audio post-processing effects. The A113X SoC boasts high performance with low power consumption, enabling application headroom for cost-sensitive OEM designs with option for battery-powered applications.
- Allwinner's SoC-Only 3-Mic Far-Field Development Kit for Amazon AVS is designed for OEMs and ODMs integrating Alexa into devices including smart speakers, table-top smart home devices with displays, home appliances, and robotic gadgets. The development kit offers far-field performance with a 3-mic array, and uses a single quad-core SoC architecture that does not require an external DSP – the combination of which delivers a compelling and cost-effective solution for integrating Alexa into finished products. The SoC supports advanced algorithms for audio echo cancellation, beamforming, and noise reduction, and provides application headroom for OEM designs with option for battery-operated designs.
Our vision at Amazon is to bring Alexa to customers everywhere - at their homes, in the office, on-the-go and everywhere in between. Voice will be woven into our daily lives and will extend beyond our Echo family of devices. As Alexa continues to be directly integrated into the brands customers know and love, we want to make it incredibly easy for developers and manufacturers to build Alexa into their products.
We continue to work with leading technology providers to bring simpler ways to build Alexa-enabled products which ultimately offers more choices to customers. See our full portfolio of qualified Development Kits for AVS.
Getting Started with AVS
The Alexa service is always getting smarter both for features, and for natural language understanding and accuracy. As a developer, your product also gains access to new capabilities with Alexa through API updates, feature launches, and custom skills. Learn how AVS can add rich voice-powered experiences to your connected products on the AVS Developer Portal.
Have questions? We’re here to help. Visit us on the AVS Forum or Alexa GitHub to speak with one of our experts.
Source: Alexa Developer Blog
