Intel® Sound Analytic Engine and Intel® Smart Home Developer Kits: Use Cases and Applications

Speech cognitive technology is all around us. From the telephone payment systems at your local utility company, to the digital personal assistant in your phone, and more recently, the smart speaker sitting in your living room, speech cognition is a pervasive and rapidly growing technology.

Over the past five years, adoption of voice-enabled devices has grown exponentially. Driven by market leaders (Amazon, Google and Microsoft) voice-controlled Smart Home devices are beginning to permeate the domestic environment. While we are most familiar with smart speakers, this is just the first phase in the evolution of the home, from a place of shelter and comfort to a valuable tool to make our lives easier. This presents an exciting opportunity for you as a product developer to add voice capabilities to new and innovative form factors. Whether it’s adding voice to a current design, or building a new, voice-first product, Intel® technology provides the building blocks for prototyping and bringing new Smart Home Solutions to market. But before determining what type of customer experience you intend to create, let’s look at some of the benefits and user requirements for adoption.

The benefits of adding speech understanding to Smart Home devices for manufacturers can be grouped into three different categories:

Devices like smart speakers allow users to access information and services online with the intuitive power of their voice. These devices typically feature Personal Assistant technology and are connected to the internet to provide Natural Language Processing (NLP) in the cloud. This allows manufacturers to deliver cloud-based services seamlessly through voice-first devices. The device providers are also able to identify and learn customer preferences, which allows them to improve over time and grow adoption. 

Speech can also be a valuable tool to understand your customers’ needs without requiring cloud-based services. Simple command and control functionality can be added to many different form factors (think coffee pots, dishwashers, microwaves) as an easy interface for users. In turn, this allows companies to better understand how their products are being used and improve the product lifecycle. 

Manufacturers looking to leverage their existing products to build high fidelity sound capable devices gain a clear competitive advantage when adding voice to their platforms. These highly-integrated, user-friendly designs often require little to no training for customers to be up and running. This allows companies to focus on continuing to develop best-in-class hardware while supporting a large application development infrastructure to bring a uniform and intuitive experiences to users.

The rapid adoption of voice-first technology in the home is due in large part to the ease and instinctiveness of communicating with your natural voice. For that reason, it’s important for developers to focus on ease of use when building voice-enabled Smart Home products. To drive user adoption, voice-enabled Smart Home devices will require low latency, low word error rate (WER), a large vocabulary (local or cloud-based), and the ability to speak and be understood from a reasonable conversation distance. 

The Intel® Sound Analytic Engine is a dual DSP and neural network accelerator that provides silicon, algorithms, and a reference design microphone array designed with complex, far-field signal processing algorithms that use high dimensional microphone arrays to do beamforming, echo cancellation, and noise reduction. This simplifies enabling speech across a range of form factors, allowing developers to add far-field voice, speech recognition, and amazing acoustics to low-power devices. It enables the user requirements by providing a building block for voice-enabling that uses a silicon-based, Intel-developed Gaussian Network Accelerator (GNA).

Intel® Sound Analytic Engine provides you with a straightforward path to developing either a cloud-based voice recognition system or a large vocabulary local speech recognition system. It allows you to bring products to market quickly with a pre-established framework for a smart speaker design that can be integrated into many different form factors. 

Intel is introducing Smart Home Developer Kits to empower hardware and software developers to quickly bring new voice-enabled products to market. The primary technology in these kits is the Intel® Sound Analytic Engine. 

The first developer kit, the Intel® Speech Enabling Developer Kit, will be available for sale in October 2017. This kit contains the Intel® Sound Analytical Engine (a dual DSP with neural network accelerator), mic arrays, speaker mount, and a Raspberry Pi* connector cable to get you quickly prototyping with Alexa* Voice Services. Future developer kits will enable additional features, including imaging and sensors.

There are two main categories of Smart Home devices that can utilize Intel® Sound Analytic Engine technology to enable speech understanding: smart appliances and smart speakers.

Transforming traditional appliances into “smart appliances” requires being able to interact with them directly. Rather than adding a keyboard and mouse or touchscreen, which still requires users to physically interact with their devices, truly smart appliances should have voice as their primary interface. This will require far-field understanding, low latency, and low power for always-on capabilities. Low cost and low power are critical for the digital microphones and speakers that power speech interaction. Adding voice to existing form factors will also require flexible designs that can fit into established chassis, like ovens, dishwashers, and tea kettles. When these requirements are satisfied, then you will achieve true value for your users. For instance, enabling speech understanding on a coffee pot would allow you to start your coffee with your voice, freeing you to accomplish other morning tasks simultaneously. 

Smart speakers enabled with Personal Assistants is a rapidly growing segment of smart home products. Research from Parks Associates suggests that adoption doubled, from 5% to 10-11% in the U.S. between 2015 and 2016. And total sales of smart speakers with Personal Assistants is estimated at 14 million units in 2016 . This rapid adoption can be attributed to the intuitive interface and utility of these devices, for everything from playing music to accessing cloud-based services for information. The Intel® Speech Enabling Developer Kit can be leveraged to prototype smart speakers equipped with cloud-based Personal Assistants. The Intel® Sound Analytic Engine provides a straightforward path to developing either a cloud-based voice recognition system or a large vocabulary local speech recognition system that provides a high-quality speech recognition experience and a broad choice of pre- and post-processing capabilities. It allows you to get to market very quickly with a pre-established framework for a smart speaker design that can be integrated into many different form factors to enable a ubiquitous speech platform.

In the last five years, we have experienced an explosion of voice-enabled devices. However, we have just scratched the surface of what is possible when you add voice in the Smart Home. From dishwashers to washing machines, speech understanding can provide a clear competitive advantage by differentiating your product from that of your competitors. Not only will the Dev Kit help to differentiate your product from your competitors, but it will give you a head-start on the future when having a personal assistant will be the household standard. It also provides a path to better understand your customers, more easily deliver cloud-based services, and maintain differentiators like best-in-class hardware. 

The Intel® Sound Analytic Engine platform can enable speech across a wide range of form factors by providing silicon, algorithms, and a reference design microphone array that is designed to enable user requirements for Smart Home devices. It uses Intel’s silicon hardened GNA (Gaussian Network Accelerator) to improve cloud and local based speech recognition, acoustic context awareness, and power reduction. This technology is available in Intel® Smart Home Developer Kits to simplify prototyping. For more information about developer kits, check out the resources below.

Links to: