Advanced Voice Control: Multiple Voice Assistants for Home Audio
In a growing and rapidly evolving home audio market — including soundbars and other speaker products — the requirement for OEMs and ODMs to produce feature-rich devices, with the latest in audio technology and a fast time to market, is increasingly only enabled through partnerships with independent software engineering houses and technology providers.
The process of building such products with extensive digital signal processing features involves the design, tuning, embedding, qualification by applicable third parties such as Amazon, and multiple stages of validation and testing. Much of this process exists in what StreamUnlimited refers to as the “development gap” between hardware selection and manufacturing.
When navigating this gap, product makers need tools that assist and streamline the processes of designing core functionality, troubleshooting, tuning, and integrating features that add value such as voice assistants and native support for streaming services.
To drive development in a competitive market, DSP Concepts and StreamUnlimited each provide sets of software technologies with similar propositions: the Audio Weaver platform and StreamSDK software stack are both modular, customizable development environments, and boast a wide range of features including third-party features and technologies. Each (and both together) enable teams to easily design the product with a portfolio of features that make it competitive, with AVS qualification and support for third-party services. StreamUnlimited also offer a range of Hardware modules, based on leading SoCs, to support the software solutions. In contrast to other AVS integrators offering basic reference designs with limited additional features and functionality, the combination of StreamSDK and Audio Weaver provides a leading-edge hardware and software solution for advanced product development.
Audio Weaver from DSP Concepts is a low-code, hardware-independent audio platform that removes the risk and complexity when designing audio products. Featuring a drag-and-drop host application and an extensive set of embedded signal processing libraries, Audio Weaver contains tools that assist product makers in streamlining their development workflow at each step, from prototype to production.
As a hardware-independent platform, Audio Weaver supports the layout and development of audio signal processing features using only AWE Designer on a PC. This enables developers to perform tuning and debugging before deploying the design.
These designs can be deployed when ready, without the need for a redesign, to a target product that has the embedded AWE Core libraries. With dynamic instantiation of the layout and its components, further debugging, real-time tuning, and profiling can then be performed using AWE Designer while connected to the hardware.
The Audio Weaver platform allows developers to work in parallel, each tuning or implementing different features independently, then combining the features into a final layout for deployment. This approach mitigates some risks associated with product design, since the audio signal processing can be designed efficiently, iterated quickly, then tested and tuned easily.
TalkTo, also from DSP Concepts, is a highly configurable audio front end (AFE) that employs advanced signal processing techniques to transmit clean voice signals to speech recognition engines and voice assistants. TalkTo can easily scale from low-power, single-microphone solutions with quiescent sound detection and single-channel noise reduction, to advanced multi-microphone arrays and a host of sophisticated signal processing algorithms such as beam forming, multichannel acoustic echo cancelation (AEC) that improves mid- and far-field voice detection for devices with a high-SPL output, adaptive interference cancellation (AIC) that reduces external noise by up to 30 dB, and more. The scalability of TalkTo makes it an ideal set of technologies for inclusion in products such as smart speakers and soundbars, televisions, smart home/IoT devices, hearables, and wearables.
StreamSDK from StreamUnlimited is a modular and customizable software solution that supports all major semiconductor suppliers and enables new development or the extension of an existing platform with streaming functionality and voice user interfaces. StreamSDK supports a wide range of third-party services and features, such as Chromecast, AirPlay 2, Qplay, a multitude of music streaming services, and playback using industry standard formats plus specialized codecs from Dolby, Sony, and more. It also supports all major voice assistants — even enabling the coexistence of multiple built-in assistants on a single device for voice interoperability across ecosystems.
For products that require an edge-based voice UI, offline voice assistant functionality is made available that utilizes Sensory™ VoiceHub for creating and implementing local, customizable wake words and command sets with no reliance on cloud computing. Another valuable aspect of StreamSDK is the ability to incorporate multiple methods of user interface, for device control via mobile app, voice, onboard touch buttons, dedicated remote control units, or web-based UI.
StreamSDK is an exemplary tool for the development of a wide range of products, from low-cost smart home devices to high-end performance audio products, due to the diversity of features and the ease of their implementation into the end product.
In a collaboration between DSP Concepts and StreamUnlimited, a reference design supporting both StreamSDK and AWE Core libraries is now available. Based on the NXP iMX8M applications processor, the Stream1955 is a compact system-on-module that features online connectivity, Dolby Atmos/Sony360/DTS:X decoding and voice assistant capability using only the SoC (no DSP required).
Qualified development kits for AVS are complete reference solutions for creating products with Alexa built in. Using this reference design, product development with Audio Weaver can be done with a unified environment for prototyping and tuning, thus de-risking and simplifying the process of creating audio layouts. The AVS-qualified TalkTo audio front end and the customized StreamUnlimited modules available from within Audio Weaver open an accelerated pathway to attaining Alexa Voice Service and Alexa communications certification, reducing complexity and time to market with a robust product offering that overhauls the competition.
As an Amazon-certified System Integrator, StreamUnlimited has natively implemented Alexa features into the StreamSDK software stack. The Stream1955 reference design is capable of Frustration-Free Setup (FFS), Multi-Room Music (MRM), Micless-MRM, Alexa Calling, Messaging, and Announcements (ACM), Amazon Music (including HD), Amazon Wake Word, and Alexa Cast.
This reference design features AVS and Google Assistant, along with the TalkTo audio front end featuring dual microphones and multichannel AEC to reduce reverberation and noise. With this audio front end, voice commands in the near- and mid-field can be executed clearly even in noisy environments.
The Stream1955 reference design from DSP Concepts and StreamUnlimited can help kick-start the design of competitive and feature-rich soundbar and smart speaker products with AVS integration, reducing complexity and time to market.
For more information about TalkTo, the Audio Weaver platform, or the Stream1955 reference design, please contact DSP Concepts.To learn more about StreamSDK, please visit StreamUnlimited.com or email firstname.lastname@example.org.