How Sonos built a more open-minded smart speaker in the era of Big Tech.

When Sonos decided to build a voice-controlled speaker, it was as much a strategic shift as it was a new product. Not only did our team face the challenge of creating a voice-enabled speaker for a multi-room sound system—something no one had yet done—but we were now going head to head with some of the biggest technology companies in the world. Thankfully, Sonos has been focused on reinventing home audio for over 15 years.

That track record of innovation and experience leadership would lend itself well to the task at hand: building a new way to listen. There are plenty of so-called smart speakers out there, but the first insight in our product development was that these remarkable devices were not actually built for listening: they use voice control and AI primarily to speed productivity, make shopping easier, and deepen engagement with certain Big Tech platforms. (And yes you could also listen to music on them.) We were determined to make a smarter speaker, one that would have the joy of listening as its primary objective.

The first and probably hardest principle we had to adhere to in order to achieve this objective was openness. Sonos has always been open to different music services, but to live up to our experience standard in this new product, we realized we would also need to be open to multiple voice assistants so people could craft their own sonic experience.

We also noticed that many smart speakers are designed in such a way that your attention goes to the tech instead of the music. The speaker had to look good of course but we always focus first and foremost on making sure our speakers fit naturally into people’s homes. That means they don’t need to be in the center of the room to sound great and they don’t need to grab your attention with a bunch of lights and alerts.

To make this ambitious vision a reality, a relatively small team banded together and dreamed up what would become Sonos One and the groundbreaking software experience behind it. Its creation, recounted here by several insiders, represents a leap forward not just for Sonos as a home sound system, but as a platform upon which new and exciting things can be built.

Built from the ground up – well, almost.

First, a voice-enabled speaker requires a voice assistant. We knew that Amazon had lots of expertise in this area. The Echo experience was inspiring and Alexa had grown into a remarkably capable voice control platform backed by some of the most sophisticated machine learning technology in the world. And rather than keeping Alexa in a closed box all to themselves, Amazon had turned it into a open platform that others can easily work with. So naturally, Alexa became our first voice control partner.

Before we could build any parts for a voice-enabled speaker, we needed to determine just what it would look like. “Ultimately we decided to build on the backbone of the Play:1,” Product Manager Allan Velzy explains, “From an experience perspective, the Play:1 was our most popular speaker at the time not only because it sounds amazing but also because it can go in any room.”

We wanted to make Sonos One visually quiet but sonically awesome.

Despite looking nearly identical, Sonos One and Play:1 only share two of the same parts: the feet and base. “Because we were adding microphones and lights, everything had to be redesigned,” says Velzy. These additional elements are a major part of a design where every millimeter counts and tweeters and woofers are tailored to the unique architecture of the speaker.

Allan Velzy, Product Manager

The team studied the Echo and thought about how to refine the Sonos One experience. “One of the more interesting challenges was we knew we wanted to offer the opportunity to turn the mics off,” he explains. They noted that when you turned the mic off on the Echo, the speaker would light up red. “It felt kind of scary and negative. We didn’t want that, so we just paired a microphone icon—no line through it—with a small LED light. If the light is off, the mic is off. We also hardwired the microphone array with the LED, so there’s no way for the microphones to be on without that LED on. It’s a little privacy feature but great for security.”

They also reviewed the Echo’s highly visual interface. For example, when you say “Alexa,” it lights up blue. Sonos took a different approach. “We wanted to make Sonos One visually quiet but sonically awesome, so we went with an audible interface,” Velzy continues. “When you say ‘Alexa,’ it makes a little chime, signaling it can hear you.”

A speaker that listens as great as it sounds.

Teaching Sonos One to hear voice commands proved to be a challenge. After all, before this project the team only had to worry about the sound coming out of speakers, not into them.

Brendan Welch, who helped prepare Sonos One for production, played a key part on the hardware side, helping to build the microphones into the speaker. “First we had to get the electrical design right so there’d be no interference with voice signals,” he explains. “Then we had to select a microphone that’s sensitive enough to pick up a voice but robust enough to handle the very high sound pressure levels output from the speaker.” Finally, the microphones had to be properly sealed with a transparent acoustic barrier to maintain consistent performance and withstand external contaminants.

Klaus Hartung, Director of Smart Audio and Voice

Klaus Hartung, Director of Smart Audio and Voice, worked with his team on software that could distinguish a voice command from music, conversations and all the other sounds in a room. “People think we can just subtract the music from the microphone signal. It is much more complicated,” says Hartung.

If we wanted to let people place the speaker wherever they wanted in a room, we had to apply echo cancellation. “Every room and every placement is unique,” explains Hartung. “Depending on the size of the room, the placement of the speaker, the materials in the room and even where people are, the sound waves change.”

Before Sonos One can ‘subtract’ the music, it must account for those changes using an algorithm, which creates a filter that simulates the acoustics of the room. Using what’s called a beamformer, the Sonos One is able to suppress unwanted sounds, such as noise from appliances. Another algorithm ensures that the speaker only reacts when it hears the famous wake word, “Alexa.”

A smart speaker that’s actually smart.

Developing voice control for a whole home sound system was something no one had ever done before, so making sure Sonos One would be able to do whatever command follows “Alexa” tested both the Sonos and Amazon Alexa teams.

“With the Echo, you can say, ‘Alexa, play The Beatles,’ and the device will start playing. We wanted the same short and sweet experience,” explains Mieko Kusano, Senior Director of Product Management. The added complication of multiple rooms, however, demanded a deeper integration with Sonos than Alexa had with any other partner. “Otherwise a command would have looked something like, ‘Alexa, tell Sonos to play the Beatles on Spotify in the bedroom.’ I think we can all agree that’s not very natural,” Kusano laughs.

Dayn Wilberding, Creative Director of Experience
Dayn Wilberding, Creative Director of Experience

The teams worked together to cut out the need to use any unnatural, excessive phrases. “Instead of ‘tell Sonos,’ you simply link Alexa to Sonos during the set up. Instead of ‘on Spotify,’ you can set your preferred music service in the Alexa app. And while you can specify ‘in the bedroom,’ you don’t have to; the speaker will assume you are speaking directly to it and any other speakers grouped with it,” explains Kusano.

 

We want the listener to enjoy their home life without tech feeling like a burden.

 

Another part of making the user experience easy and intuitive was implementing continuity of control. That means that however the listener chooses to control Sonos–through our app, via partner apps like Spotify, the buttons on the speaker, or by using their voice–it should work flawlessly, even when people jump from one control mechanism to another.
“We want the listener to enjoy their home life without tech feeling like a burden,” says Dayn Wilberding, Creative Director of Experience at Sonos. “People shouldn’t have to think about how music started or determine the ‘right’ interface for a task. Everything should just stay in sync.”

Even better things to come.

Lidiane Jones, Senior Director of Software Product Management

Sonos One was designed not only to work every time but also to get better with time. That’s why we built our own platform to manage the voice experience, making sure it could keep up with the rapid pace of technological change. Certainly, Alexa will get smarter over time–But so too will the entire Sonos platform, as new AI-powered voice assistants are added.

“Voice control is still in its infancy,” explains Lidiane Jones, Senior Director of Software Product Management. “We built Sonos One with the flexibility to innovate and evolve. Over time, with updates, customers will get support for Google, voice control for an expanded set of music services and additional features in the Sonos app. We knew we wanted to support multiple voice services the same way we support multiple music services.”

For a relatively small company that started by the beach in Santa Barbara, and that still operates outside the orbit of Silicon Valley, building a product as complex as Sonos One in a short time frame came with some inherent and unprecedented risks. Most daunting of all was the possibility that the user experience wouldn’t match the high expectations of customers. After wrapping up many months of concerted, almost frantic effort, the team collectively exhaled and awaited the response.

“Voice lets the technology fade away…”

Michael Papish, Director of Platform Marketing
Michael Papish, Director of Platform Marketing

“The best-sounding smart speaker you can buy,” declared one reputable technology news site. “Alexa gets the sound quality she deserves,” wrote another. Nearly across the board, the consensus started to emerge: Sonos One managed to combine voice control, flexibility and sound quality in a way that no one else had.

For Sonos, this achievement wouldn’t have been possible without an unprecedented effort on the part of multiple teams across the organization, all of which flowed from the moment that they realized just how powerful voice could be.

“It was like magic,” Michael Papish, Director of Platform Marketing recalls thinking of voice control when the project began. Now there’s less mystique surrounding the technology, but that sense of wonder remains unshaken. “But I guess the real magic was in the teamwork and focus on the customer coming together in this technology that will keep getting better until it feels like it just disappears,” Papish continues. “We’re in a new world.”

Also available in: en-gb en-au fr de it es da nl sv no