Businesses’ Voice Assistants Are Coming to a Smart Speaker Near You

“I’d lift the handset off the black phone and tell the telephone operator ‘Call Bob’s.’ The operator asked, ‘Bob’s Meat Market or Bob’s Auto Service?’ and I would respond, ‘Auto.’ Then, the operator would connect me to Bob’s Auto Service, and I’d start speaking with Bob.” My grandfather, who had one of the first telephones in his small Wyoming community, often told this story by concluding, “Connecting to a business was simpler then.”

The name of a business is one of its biggest assets. Names appear on business websites and social media sites, and increasingly they serve as the activation word—the word spoken by the customer to invoke a business’s voice agent. And just as every business has a website today, in a few years every business will have its own voice agent.

Previously, to find information about a business, customers searched the White or Yellow Pages of a telephone book (remember those?), then placed a call to that business. The internet changed that. Now customers Google the business name, wade through related or similar website descriptions, select a website, and click. Up pops the business’s web page. But there should be a faster, easier way to connect to a business than searching, scanning, and selecting—like speaking the name of a business to invoke the business voice agent.

Call Them by Their Name

Connecting with a business voice agent through your home’s smart speaker seems like a natural evolution. But there are problems:

• Some businesses have names that are pronounced the same, but have different spellings, such as Cisco (communications hardware) and Sysco (food services for restaurants).

• Other businesses have names that are pronounced and spelled the same, such as Bob’s (Meat Market) or Bob’s (Auto Service).

• A business might have multiple names, such as Gone Fishing, Went Fishing, and Go Fishing.

• A business could have a name that is not easily pronounced by English speakers, such as businesses with non-English names such as Huawei or Baidu.

Additional information—such as the customer’s history, local context, an ontology of names, or perhaps a conversation with the customer—might be needed to resolve these problems.

In Search of a Voice Registry System

A Voice Registry System (VRS) is akin to a Doman Name Service (DNS) for voice agent names. After a voice agent is registered with the VRS, no one else may register and use that specific voice agent name. When the customer speaks a voice agent name, the VRS will invoke the appropriate voice agent by locating the voice agent address. If there are ambiguities, the VRS attempts to solve them with local information or by talking with the customer. Customers can quickly connect to the business by speaking its name, just as my grandfather did years ago.

Unfortunately, a global VRS does not exist yet, but the good news is that the Open Voice Network (OVN) is developing one. If you would like to follow OVN’s progress or join the conversation with any of OVN’s standardization efforts, please visit https://openvoicenetwork.org/join/.

IVA Conference Calls

A customer can connect and disconnect sequentially with several voice agents, one at a time. But customers might want to converse with several agents at the same time—for example, when comparison shopping or coordinating appointments. Think of it as a conference call rather than several sequential individual ones.

Amazon’s Voice Interoperability Initiative (VII) recently published a multi-agent design guide to help customers interact with multiple voice agents at once. Customers start by speaking several voice agent names; the agents can interact only with the customer, not with each other. It will be an adjustment; some customers might find it confusing having a group call with several voice agents.

Take Away

Just as all businesses have web pages that can be searched and viewed by customers, soon all businesses will have voice agents that help customers locate needed information quickly.

Get started by implementing a simple voice agent, perhaps with the FAQs from your website. Stake your claim on your voice agent name by creating, registering, and installing versions of your voice agent on each of the major voice agent platforms (Amazon, Google, Samsung, and Apple). If you haven’t already done so, consider copyrighting or registering your voice agent name with the copyright office in your country.

James A. Larson is program co-chair of SpeechTEK 2021 and senior technical adviser with the Open Voice Network. He can be reached at jim42@larson-tech.com.