Multiple voice assistants on one product

By: Karen Scates

Oct 08, 2019

Alexa…I Need a Wake Word for My Product

There are a lot of predictions about the future of voice assistants. Industry experts forecast the integration of voice technology into nearly every application in some shape or form over the next 5 years, but building this technology in-house may take years without internal expertise.

Instead, leading brands, like Mercedes-Benz and Pandora, are partnering with Voice AI experts to develop custom voice user interfaces that have the flexibility to include Alexa and Google alongside a custom, branded voice assistant. These partnerships are creating voice assistants in less time and with greater success than trying to build the solution from the ground up with internal teams.

Smart speakers reached “critical mass” at the end of 2018 when 41% of US consumers were reported to own a voice-activated speaker and Alexa led the market. Consumer’s comfort with the Alexa interface and functionality has driven many brands to move toward Alexa integration into their products, sacrificing brand awareness for the comfort and ease of offering a familiar voice interface for their customers.

What many of these brands don’t realize is that they can have both: A unique, custom voice assistant in combination with the familiarity, comfort and functionality of Alexa. It allows for multiple voice assistants to live side-by-side and is enabled by a multi-phrase strategy.

Implementing a multi-phrase strategy

Creating exceptional customer experiences with your voice interface may include providing access to more than one voice assistant. For instance, you can create a branded wake word like, “Hey Mercedes” and have that voice assistant help with in-car entertainment and product-specific functions, such as rolling down the windows or regulating the temperature. In addition, you may want to integrate Alexa and allow your customers to make Amazon purchases using their voices.

To achieve this multi-phrase strategy, developers will need to partner with a voice user interface platform that includes a flexible phrase spotter. Using this technology, brands can create multiple wake phrases in one model (importantly, with minimal CPU impact to avoid burdening local systems). Once implemented, users of the product can trigger their voice assistant of choice without interrupting the flow of the conversation.

One of the great challenges of implementing a voice strategy is determining your brand’s voice. Implementing a multi-voice strategy helps to solve some of the issues associated with creating a custom voice experience that’s branded, while simultaneously offering voice-based services familiar to your audience.

“Alexa” resident on your product

You wouldn’t allow another brand to determine your logo design and colors, why would you give away your core voice experience? While voice assistant adoption has not yet reached saturation in all applications, it soon will. Rising above the noise is already difficult for most brands.

The advent of voice in all products that require a human-to-machine interface will further deepen the challenges of developing a unique and memorable brand identity. To stay competitive, innovative companies will consider ways to differentiate products through voice and prove value beyond that which is promised by the competition.

“With many voice assistants, we lose an important component…sight, and rely mainly on hearing as a means to spark association, memory, and interaction.”

— Elissa Daly, Senior Strategist, RAIN.

In a world where brand ambassadors play a strong role, we search for ways to create better experiences and build branded emotional connections with customers. When we can provide personalized experiences, we inch closer to developing the kind of brand loyalty that ensures our products will outlive the next clever advertising campaign or product feature introduced by competing brands.

As voice interfaces move from being a “nice to have” to a critical component of your product, you’ll need a way to make sure you don’t sound like just another product in a crowded space. When everyone is talking, you’ll want to make your customers’ experiences with your brand is recognizable and memorable. As part of that challenge, the interactions with your voice interface must add a level of convenience and allow for conversational exchanges between your users and the machine.

Conversational interactions with voice assistants

When we ask a series of questions of someone who is sitting in the same room as us, we don’t often use that person’s name at the beginning of every query. A key element of creating conversational exchanges between people and devices is the ability for the devices to allow for everyday actions without always requiring a specific wake word trigger.

Locally embedded command recognizers allow devices to do just that. When a user asks for the same information time and time again, the command recognizer can respond to the common phrase without the need to hear the wake word first.

When that command recognizer is embedded into the hardware, response times are shortened — making the conversation even more natural. Everyday commands become increasingly convenient and a natural extension of the product experience.

For example, telling a washing machine to “start the wash cycle” is a command very specific to that device. Not requiring the standard wake word creates a seamless experience. Therefore, a company’s wake word partner needs to offer this capability in anticipation of the evolving nature of conversational interfaces.

Furthermore, a voice assistant that can adapt to the user’s personal context is critical to establishing a personal connection between humans and machines. Therefore, identifying the speaker’s voice allows for individual preferences to be accounted for in the delivery of a result. With repeated behavior, the system can make recommendations based on patterns of behavior for that individual.

“Depending on where you’re going, who you’re calling, what radio station you’re listening to in the car, the system will learn that. So, the next time that driver is in the car, it will recommend those choices.”

— Mihail Antonescu Sr. Software Engineer, Mercedes-Benz R&D North America.

Of course, you’ll want to guard against false positives that result as part of environmental and background conversation. Having the voice assistant wake up when hearing words similar to the wake phrase does not result in a positive experience for the user — and may lead to frustration with the device. Adding in a level of rigor around the testing phase of the voice assistant will help reduce the incidence of false positives.

Shortcut solutions that don’t include multiple levels of testing are likely to end with voice interfaces that frustrate the user and result in poor product experiences.

Finding the balance between conversational interactions that happen quickly and naturally, and developing a device that either doesn’t wake up or wakes up too often requires a blend of technologies proven to provide the best experience possible.

Building an Alexa-certified wake word

Regardless of the voice strategy you choose: a branded wake word only; a branded wake word alongside Alexa or other assistants; or choosing Alexa only — make sure your voice assistant is delivering real value and enhanced product experiences for your users.

If you decide to include Alexa in your voice assistant experience, you’ll want to make sure you’re working with a partner who can build an Alexa-Certified wake word. The certification ensures your custom branded voice assistant will work seamlessly with Alexa and deliver the best user experience possible.

Houndify’s independent platform allows for full freedom of brand ownership, user experience and customer data. Our mission to “Houndify Everything” includes building custom voice assistants that include a multi-phrase strategy and allow other assistants to live alongside your branded experience — providing complete control over your customer experience and data.

Find out how to align your voice strategy with your brand strategy with Houndify.

Karen Scates is a storyteller with a passion for helping others through content. Argentine tango, good books and great wine round out Karen’s interests.

Interested in Learning More?

Subscribe today to stay informed and get regular updates from SoundHound Inc.

Subscription Form Horizontal