Consumers already expect that in-car voice assistants will be responsive and accurate all the time. Whether or not that is happening is a point of disagreement between consumers and car manufacturers. Capgemini reports that 81% of automakers believe their embedded voice assistants are meeting consumer’s needs, while only 59% of consumers agree.

The growing consumer demand for voice assistants in-car represents a massive opportunity for automakers willing to continue iterating and improving the voice user interfaces embedded in their vehicles. In the future, consumers will expect more out of voice assistants, according to the report.

As a provider of an advanced voice AI platform, Houndify, SoundHound, Inc is already working with top auto manufacturers to develop voice assistants that meet these three key criteria:

1. Responsive and conversational
2. Contextual
3. Connected

Beyond ASR and NLU

According to Capgemini, half of the users of voice assistants in cars feel that they are not accurately understood. Although the voice industry is moving toward greater natural language understanding, most voice assistants still require a two-step process for converting speech to data and back to speech.

Automatic Speech Recognition (ASR) technology allows people to talk to machines and for machines to understand the queries. In a second step, machines will make sense of the data and reply back using Natural Language Understanding (NLU). The process takes time and there is still room for error.

What people really expect from their voice assistants is a conversation that happens as naturally as talking to another person. This level of intelligent conversation is available through a breakthrough technology referred to as Speech-to-Meaning.

An early advancement in the voice first world, Speech-to-Meaning technology is able to understand the meaning of user queries as they are spoken. Because the two technologies, ASR and NLU, have been combined into one function, users can ask questions and get responses from their voice assistants at the speed of thought.

Within three years, Capgemini predicts that 73% of people will be relying on voice interfaces — an increase of 24% from today. Those same consumers will be looking for exceptional experiences in one of the two places they most commonly use voice today — their cars (The other being the home).

Addressing the most common in-car use cases

Both Capgemini and SoundHound Inc. have identified playing music, navigation, and making calls as the three main uses for in-car voice assistants today. Those use cases are predicted to grow exponentially in the next three years — and will increasingly combine the functionalities of in-car and at-home assistants.

Building voice assistants that satisfy consumer demand for seamless user experiences means ensuring that all the functions available through the voice assistant work in the context of driving. Designing a voice assistant for the car environment requires taking into consideration the various elements of being in a vehicle, including the role of road noise and other people in the car. Automakers will also want their vehicles to offer interfaces for the most common languages and for the subtleties of language within a country market where many cultures live side-by-side.

Despite the temptation to design a voice assistant that performs unusual functions — such as delivering sports scores or giving the day’s stock market results — experts agree that getting the basics right is the best path to brand stickiness and user satisfaction.

The challenge for automakers is that the life of their cars will outlast the growth of the available technology. Today, delivering the level of user experience consumers expect is the greatest challenge. Tomorrow, having access to more types of functions and a growing list of domains will be some of the factors that will help individual auto manufacturers remain competitive.

Building the right voice interface today for greater functionality tomorrow means looking to cloud-based solutions that allow manufacturers to update technology, retain customers, and build brand affinity over time. Collective AI™ is an example of a cloud-based, crowdsourced domain platform that is always learning and expanding — giving brands a seamless avenue to continue expanding the functionality of their voice assistants.

Universal connectivity with voice AI

“Consumers want to port their home-based voice experiences into the car,” according to the report by Capgemini. It’s clear that car makers who can make those kinds of connections are likely to grow market share. What isn’t clear is how to provide a high-level of connectivity — now in demand from consumers — without giving away valuable customer connections.

If automakers create partnerships with existing providers, such as Google and Alexa, they are in danger of losing the ability to monetize the digital activities that happen within the car. However, not including the voice interfaces most familiar to their consumers means possibly losing those drivers who love interacting with their in-home voice assistants. In the current voice AI environment, there is no obvious choice that guarantees the least amount of risk.

Fortunately, there is an alternative that provides the functions and familiarity of Google or Alexa, but leaves the control of data and customer connections in the hands of the auto manufacturer. A multiple assistant solution allows for a custom voice assistant and existing voice assistants to live side-by-side in the same vehicle — by leveraging more than one wake word—known as a multi-phrase strategy. Car manufacturers embracing a combination of both custom “owned” and third-party voice assistants have the opportunity to deliver unique in-car voice experiences that satisfy their user’s needs for the familiar without giving away the brand and data.

Making connectivity possible

Voice assistants that are embedded in the vehicle will never be able to offer the kind of universal connectivity of the future. In addition, cloud-based only voice assistant solutions will not be able to deliver the consistent levels of always-available service consumers are demanding. The way forward is clear: A hybrid solution of embedded voice technology plus cloud connectivity.

A hybrid (or dual) solution ensures that the voice assistant is available for the most basic functions even when the car is out of range. A hybrid approach also makes it possible for car manufacturers to plan for greater connectivity between their cars and home devices in the future — even if those functions are not available today.

For most car manufacturers, autonomous cars are on the future roadmap — and that future is getting closer all the time. For leading automakers, connectivity between the home and the car will soon be a must-have. Brands hoping to compete will follow the lead of industry innovators to put the best technology in place now as a defense against the future.