Although voice AI has been around for nearly a decade, it's now starting to unlock new opportunities for manufacturers and service providers across industries. A broader range of connectivity choices has been key to removing barriers to entry— including cost, privacy, processing power, and time to market.
Many brands have recognized voice AI as a solution to their business challenges and are enthusiastically adding it to their product roadmaps, but it’s critical they get it right and deploy voice tech with the right capabilities to meet the needs of their users.
If you’re one of them, here’s what you need to know:
Cloud connectivity
When we talk about voice assistants in the cloud, we are referring to those voice user interfaces built for the purpose of retrieving up-to-date information and accessing functionality from content domains available through an internet connection. Voice assistants with cloud connectivity are typically accessed via speakers, cars, and anywhere that users may need instant answers to a variety of questions.
Most well-known voice assistants operate predominantly, or exclusively, using the cloud.
Examples:
- “What’s the weather like in San Jose?”
- “Is my flight on time?”
- “What’s the latest score?”
Edge voice assistants
When voice AI has little or no connection to the cloud, it is considered to be on the edge. Even without a cloud connection, these voice user interfaces can perform a wide range of functions, provide hands-free access to devices, and deliver a limited set of information to the user.
Edge voice assistants provide the greatest level of privacy as information is stored on-device and isn’t accessible from the cloud (something that can be critical to consumer trust). Depending on the use case and manufacturer needs, edge voice assistants can also include as much or as little Natural Language Understanding (NLU) functionality as their solutions demand.
When the cloud is not available, privacy is important, or device processing power and costs are limited, intelligent edge voice AI technology is giving brands across industries a voice. It can work especially well for smaller smart devices, or those with a limited number of specific use cases. Smaller footprint edge and hybrid voice assistants are emerging as solutions for smart devices, as well as hearables, wearables, wireless headphones, and a number of devices in the medical and manufacturing industries.
Examples:
- “Send a text message to Marie.”
- “Turn off the lights.”
- “Increase the temperature by 2 degrees”
Hybrid voice solutions
In some instances, manufacturers and service providers want their users to have access to the cloud to ensure a consistent voice experience. Hybrid voice assistants provide the most device control, personalization, and information gathering capabilities of any other solution.
Where device capacity is not an issue OEMs, like automotive brands, can deliver everything from device control to stock prices and flight status to their customers. Where the device has less capacity available, they can use cloud connectivity to answer some queries or push product updates without gathering usage data—thereby combining the privacy of embedded technology with the benefits of product and voice optimization over time.
Hybrid solutions enable car manufacturers and other OEMs that don't want to rely exclusively on cloud connectivity for hands-free operations of their hardware to ensure always-on service for their customers. But moving vehicles aren’t the only use case. In fact, we all experience poor internet connections—even at home. And relying solely on the cloud for voice assistant operability may not be the best option when voice commands are only really used to turn on and off the product or determine specific settings.
Indeed, where products and services have a mix of capabilities, they aren’t all best served by the cloud. With hybrid voice solutions, functions that don’t require cloud connectivity can easily be served locally without information traveling to the cloud unnecessarily, leading to faster responses.
Edge and hybrid solutions also have the advantage of allowing users to operate devices and get a pre-set number of answers to questions for things like directions to use the product and maintenance records when the cloud is unavailable, unreliable, or not required.
Examples:
- “Start a wash cycle and place an order for more detergent”
- “Turn on air conditioning and navigate me to the nearest salon”
- “Play music and place our usual order with Pizza Guys”
Improving technology
In the past, hybrid voice assistants could be slow to respond and difficult to access. In some cases, the technology for the embedded assistant was not the same as the cloud-connected voice AI. In other cases, the arbitration software that allowed the voice assistant to access both the embedded content and domains in the cloud, treated the different content locations as separate entities—waiting for a possible response from the embedded voice assistant before moving the query to the cloud (or vice versa).
Now, parallel arbitration models provide faster and more accurate voice experiences by sending queries and requests to both the edge device and the cloud voice AI simultaneously in a hybrid solution. Whichever system provides the fastest, most accurate answer is the one that is delivered to the user. Parallel arbitration eliminates the delay in response common in legacy technology solutions and delivers a seamless experience to users who are unaware of the source of the response
Getting it right
For brands and manufacturers, understanding the end-user, knowing their environments, needs, and desires are critical to deciding on which option is relevant or necessary. The great thing is that increased choice is allowing manufacturers to find fast, accurate technology that’s the best fit for their needs—and in some cases making quality voice control accessible for the first time.
With consumers embracing voice technology as the next interface, it’s encouraging that there’s ample opportunity for any manufacturer and service provider to create an exceptional user experience.
