Qualified AVS Solutions include chipsets, AFE technologies and modules from ChipSet Vendors (CSV), Algorithm Providers (ALGOP) and module makers. Qualified solutions (which include chipsets and voice processing technologies packaged with client software) enable developers to speed up the development process and lower costs for creating commercial-grade ABI products.
The qualified chipset-based AVS device solutions include three categories, non-display AVS, display AVS and hearable solutions.
Non-display solutions are utilized to build smart speakers, soundbars, set-top boxes, etc. These solutions comprise microphones, speakers, and three processing blocks: AFE, WWE and AVS Device SDK. Alexa Calling, Messaging (ACM), Frustration Free Setup (FFS), and Multi-Room Music (MRM) are add-on features. The non-display solutions can be implemented with either a single-chip SoC architecture or a combination of SoC and an external DSP/MCU.
The block design shown here represents a single-chip SoC for non-display AVS solutions. All components run on the SoC, which must be security-approved. For qualification, AFE, WWE, and AVS client with AVS Device SDK are mandatory. Add-on features like ACM, FFS, and MRM are optional.
This is a block diagram of a dual-chip non-display solution that combines a SoC with an external DSP/MCU. AFE algorithm runs on the DSP/MCU, while the WWE can be on either the SoC or DSP/MCU. As the AVS client is running on the SoC, it must be security-approved. Similar to a single-chip SoC solution, AFE, WWE, and AVS client with AVS Device SDK are mandatory components for qualification. Add-on features such as ACM, FFS, and MRM are optional.
Display solutions are used for creating smart displays, TVs, set top boxes, etc. They include microphones, speakers, and three main processing blocks (AFE, WWE and AVS Device SDK), with add-on features like ACM, FFS, and MRM. In addition, the solution requires Alexa Smart Screen Web Components for full implementation of Alexa's multimodal features.
Display solutions integrate the AVS Device SDK with Alexa Smart Screen Web Components, using either a single-chip SoC architecture or a dual-chip architecture that combines a SoC with an external DSP/MCU.
The block design illustrated here is a typical example of a single-chip SoC for display AVS solution. All components run on the SoC, which must be approved for security. Essential components for qualification include the AFE, WWE, and the AVS client with AVS Device SDK and Alexa smart screen web components. Add-on features like ACM, FFS, and MRM are optional.
This block diagram illustrates a dual-chip display solution that combines a SoC with an external DSP/MCU. The AFE algorithm is running on DSP/MCU while the WWE can be on either SoC or DSP/MCU. The AVS client is running on the SoC, which must be approved for security. For qualification, the mandatory components include the AFE, WWE, and AVS client with the AVS Device SDK and Alexa smart screen web components. Add-on features like ACM, FFS, and MRM are optional.
Hearable solutions are developed for Bluetooth devices such as headphones, headsets, and earphones to provide full voice interaction with Alexa. By integrating the AMA kit, your Bluetooth-capable device can become an Alexa endpoint by connecting directly with the Amazon Alexa app on a mobile device.
This is a block diagram of a single-chip SoC/MCU for hearable solution. All components are running on the Bluetooth SoC/MCU, which must be approved for security. For qualification, AFE and AMA protocol components are mandatory. The WWE is only mandatory for voice-initiated hearable solutions.
This is a block diagram of a dual-chip hearable solution that combines a SoC/MCU with an external DSP solution. The AFE algorithm is running on the DSP. Since the AMA protocol is running on the Bluetooth SoC/MCU, the SoC/MCU requires security approval. For qualification, the AFE and AMA protocol are mandatory. The WWE is only mandatory for voice-initiated hearable solutions.
The Audio Front End solution is the combination of multiple technologies created to improve the signal quality of an audio source, ensuring the “Alexa” wake word detection in both quiet and noisy environments. There are two types of AFE solutions, software IP AFE and silicon-based AFE solutions.
For software IP AFE solutions, the AFE algorithms can run on different DSPs, MCUs or SoCs. AFE solution providers only offer SW IP algorithms, such as Noise Suppression (NS), Acoustic Echo Cancellation (AEC), and Beamforming.
The silicon-based AFE solutions contain DSP or MCU ICs with built-in AFE algorithm and WWE. AFE solution providers offer both the ICs and Software IP algorithms, such as NS, AEC, Beamforming, even the qualified WWE.
This is a block diagram of a software IP AFE solution on SoC. The AFE Algorithm runs on the SoC. For qualification, only the AFE is mandatory. The AVS client and the WWE are part of test setup.
This is a block diagram of a silicon-based AFE solution. It contains DSP/MCU ICs with integrated AFE algorithms. The WWE can be on DSP/MCU or SoC. If the WWE is running on the DSP/MCU, both the AFE and WWE are mandatory for qualification. If the WWE is on the SoC, only the AFE is mandatory. The SoC with AVS client is part of test setup. However, if you choose this type of AFE solution for an ABI product, the developer must confirm the readiness of the WWE on the selected SoC or DSP/MCU.