Custom Voice Assistant
Jan 28, 2021

What Makes a Custom Voice Assistant Valuable?

Recently, the lines have been blurred between a custom voice assistant, a branded voice assistant, and a wholly-owned voice assistant—terms that were used interchangeably in the past. Ongoing evolution in the voice AI industry has been exciting, but has introduced complexities, making it worth taking a closer look at the unique properties, benefits, and value of a custom, branded voice assistant. 

Key considerations for a custom voice assistant

As voice AI becomes ubiquitous, companies with voice-first strategies on their roadmaps have recognized the need to differentiate themselves from their competitors with a single voice across all channels—from mobile apps and customer contact centers to websites and hardware integrations. This alignment of voice-related experiences represents a true breakthrough–a single “voice” for a company across hardware, software, mobile apps, and call centers.

These same organizations have further realized that sonic branding is a key element of their future reputation and position in the market. 

In the last three years, top companies have moved past adding a skill into an existing voice AI platform to creating strategic voice implementations. For top companies, finding the right voice AI platform provider has been key to these efforts.

Specifically, they are looking for advanced voice technology that delivers fast, accurate answers in response to natural language questions—even in the presence of  imprecise speech, heavy accents, and noisy environments, In addition, global brands have identified a need for the flexibility to offer their voice assistants in languages that match those spoken by their users. 

In the last three years, top companies have moved past adding a skill into an existing voice AI platform to creating strategic voice implementations.

Regardless of how many functions can be added to a generalized voice AI platform, the underlying technology that powers the platform determines how well it performs. Voice AI technology that was accurate only 75% of the time 3 months ago, isn’t delivering better performance today just because additional commands or a wake word has been added to it. 

Owning the user experience and data

Just because a voice assistant platform can be modified to allow certain customizations, such as a wake word or specific commands and information, doesn’t mean that it provides the two most critical aspects of a branded voice assistant: Ownership of data and control over the customer relationship. These two key differentiators are what originally spurred the rising adoption of  custom voice assistant solutions and they remain the two critical aspects of a brand-owned voice strategy today.

Those companies that have already launched their custom “brand-owned” voice assistants—Pandora, Mercedes-Benz, Bank of America—have made the investment because they recognize the imperative to control the user experience and retain valuable data. 

Controlling costs

Implementing a custom voice assistant that meets, and even exceeds, customer expectations requires a strategic investment in time and resources. Taking shortcuts to save time and money up front is likely to result in higher costs and limited upside in the long-term. Whether the impact on the bottom line appears as a direct result of the customized third-party owned voice assistant or not, the loss of opportunity will be felt in lack of user data and lackluster brand loyalty.

Leading car manufacturers were some of the first to realize that adding a third-party voice assistant to their vehicles could cause costly and unexpected hardware and software upgrades—should the voice AI provider decide to update their technology. In response, top automakers have opted for custom voice assistant solutions that give them complete control over their brand experience.

Taking shortcuts to save time and money up front is likely to result in higher costs and limited upside in the long-term.

In the future, car manufacturers—and others—adopting a third-party branded voice assistant may find future limitations in their ability to meet growing demands of drivers, nurture brand affinity, or inform the product roadmap—costing them future sales and competitive advantages of a wholly-owned voice interface.

As a result, the largest organizations in the world are seeking voice AI platform providers to work side by side with internal teams to create branded voice assistants in considerably less time than building it themselves.

What is a custom voice assistant?

Before embarking on a voice-first initiative, take a step back and ask yourself and your stakeholders “why” you are voice-enabling your hardware, customer center, communications system, or mobile app. If the answer is to simply not get left behind, it’s tempting to choose a third-party assistant. If, on the other hand, your goal is to create unique user experiences, grow brand affinity and loyalty, and create an omni-channel voice experience, then you might want to carefully consider how you move forward with your initiative. 

Key elements of an exceptional branded voice experience include:

Custom wake word

An artfully created custom wake word takes into consideration a variety of factors. In addition to making it easily pronounceable for your intended user, wake word design must reduce the incidence of false positives and negatives, avoid certain rhymes or other negative associations, factor in environmental conditions, and be designed with the future of the brand in mind.

As your product and business matures, you’ll want to make sure your custom voice assistant and wake word are key elements in your overall corporate roadmap and not an afterthought that detracts from your brand.

Proven TTS for unique voices

Having the flexibility to choose a voice to represent your brand is critical to your sonic branding image. However, you will want that voice to be tailored to your audience, beyond simply selecting the sound. The TTS technology behind the voice should be customizable to include multiple languages, specific pronunciations in your brand lexicon, and the ability to add emotions to your voice assistant.

Custom voices built with advanced TTS technology improve the naturalness, intelligibility, comprehensibility, and intonation of the voice assistant. A voice without customization is just another voice and not part of a truly custom voice experience. 

Custom commands and broad content domains

Creating a unique user experience begins with designing a specialized voice assistant with deep knowledge of all domains most relevant to the individual brand. Companies that have already implemented successful custom voice assistants realized early on that people aren’t asking Pandora for the weather, nor are they asking Bank of America’s “Erica” to set an alarm. Users don’t need branded voice assistants to offer every functionality—and they don’t expect it. What they do expect is a voice assistant that can answer questions spoken without needing to know a specific script—naturally, the way we talk to each other.

In the future, we won’t be looking at one voice assistant to rule them all, instead the ubiquity of voice assistants will require specialization. Just as we don’t expect our family doctor to be our source of information on financial investments, our future conversations with voice assistants will be determined by their areas of expertise.

In the future, we won’t be looking at one voice assistant to rule them all, instead the ubiquity of voice assistants will require specialization.

Multiple language capability

Companies with global reach have four options:

  • Hope that all your customers speak the language of your assistant 
  • Develop an individual voice assistant for each country
  • Train an assistant to be multilingual
  • Partner with an established voice AI platform with multilingual capabilities

As companies develop their voice-first strategies, taking a long-term view will save time, resources, and money. Settling for a voice assistant solution that allows users to speak two predetermined languages is not the same as a custom voice assistant that can speak the languages and dialects of your users.

Even within a language, there are dialects and accents that can interfere with a seamless voice experience. Companies with product distribution in multiple countries will need voice assistants that are localized to respond in a culturally acceptable manner. These customized assistants will understand speech in the context of the user, including the ability to recognize and respond to the differing grammar rules and slang expressions of specific regions.

Customization has never been one-size-fits-all. Global organizations with an eye to the future will want to carefully consider their future roadmaps before adopting any voice AI platform that doesn’t already support multiple languages with the capability to add more as needed.

Customization has never been one-size-fits-all.

Embedded, cloud, and hybrid connectivity

There are many ways for companies to approach a voice-first strategy. Don’t assume that every voice assistant has to be able to answer questions on a range of subjects, including weather, local search, stock prices, sports scores, news, restaurants, and parking information. In a world of customized voice assistants, there’s no need to remain tethered to the cloud or to a third-party platform in order to deliver a convenient and hands-free experience.

Depending on the level of functionality required from the voice interface and the amount of memory and processing power available, companies can choose from a variety of connectivity options, including:

  • Embedded: No cloud connectivity. The voice assistant is resident on the product and serves to improve product functionality.
  • Cloud-Only: 100% connection to the cloud. The voice assistant has the ability to access a wide range of types of information.
  • Hybrid: A combination of cloud and embedded connectivity. The voice assistant has the ability to process the user’s query in the offline mode while simultaneously sending the query to the cloud—and responding with the best response available.

Customizable voice AI solutions offer a range of connectivity choices and give brands control over how much of their user’s information is stored in the cloud and how much of it remains resident on the device itself. Since security and privacy are trigger points for consumers, the ability to protect customer information and communicate privacy policies will continue to be key factors in the future acceptance of voice assistants. 

In a world of customized voice assistants, there’s no need to remain tethered to the cloud or to a third-party platform in order to deliver a convenient and hands-free experience

Multi-assistant solutions

The concept of a multi-assistant solution is not new. In the most common model, a custom voice assistant is resident in the same device as another assistant, such as Alexa, to allow users to make purchases or access saved preferences. Using advanced technology—such as Houndify’s Flexible Phrase Spotter—brands can create multiple wake phrases in one model with minimum CPU impact. This configuration allows users to trigger the voice assistant of choice without interrupting the flow of the conversation.

The many benefits of implementing a multi-assistant solution are unrealized if both assistants are an extension of the same voice AI platform. If your custom voice assistant lives on top of a platform you don’t fully control, you won’t truly have access to your valuable data or your customer relationships, and you won’t own your voice roadmap.

Offering your customers the comfort, ease, and familiarity of a familiar voice interface becomes a more powerful strategy when coupled with brand awareness and sovereignty.

The many benefits of implementing a multi-assistant solution are unrealized if both assistants are an extension of the same voice AI platform.

Engineering expertise and support

Without the support of a team of voice AI experts on the platform side, companies are left to do the heavy lifting of voice AI design and development on their own. Although many organizations are making an investment in voice AI infrastructure, few have the resources to design, develop, and test a truly customized voice assistant.. 

True customization isn’t easy and it isn’t fast, but it’s an investment in a world where people will soon be interacting with the technology around them the way they interact with each other—by talking. 

Either your hardware, contact centers, and mobile apps will be equipped with a voice assistant that can grow with our team, or your brand will become just another skill on a unwieldy platform filled with skills from every source imaginable. 

SoundHound Inc has been providing custom voice solutions to top brands through our Houndify advanced voice AI platform—built on technology 15 years in the making. Contact us to find out how we can bring your voice strategy to life. Developers can explore Houndify independent voice AI platform and register for a free account.

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