When Google introduced its $399 Google Home Max speaker at its press event in San Francisco on Wednesday, the company didn’t just reveal a bigger, better-sounding speaker. The device is also a first step to use Google’s vast computing resources to make sound itself smart.
Google has been dabbling in home audio for some time. It teamed up with consumer electronics companies like LG and Sony in 2015 to bring its Chromecast technology to a range of speakers. This year, it doubled down on those efforts by bringing the Google Assistant to third-party speakers, including pricier models that aim to appeal to the ears of audiophiles. So why did Google decide to also build its own high-end smart speaker?
“We are only gonna do (a project like this) if we feel like we can really innovate,” said Google VP Rishi Chandra, who leads the company’s home hardware efforts, in an interview with Variety on Wednesday.
Last year, Google introduced the original Google Home to do just that in the area of voice control and assistance without the presence of a display. This time around, it didn’t just want to add the Assistant to a bigger speaker, but actually use technology to improve sound beyond the specs of the device itself.
That’s why Google Home Max packs six microphones, instead the two that are integrated into the original Google Home and the also-newly-announced Google Home Mini. These six microphones monitor the music playing in a room for subtle acoustic changes. The speaker then senses the modalities of its environment — whether it is placed on a shelf that adds vibration sounds every time the bass line kicks in, or whether it is in a sparse room without much furniture to absorb any sound, leading to echoing of higher frequencies — and adjusts the sound output accordingly.
And these adjustments aren’t just one-offs. Google has been training the Google Home Max in the wild on thousands of rooms to develop models of these environments and their acoustic qualities, and is using this training data to tweak the speaker’s output in real-time. A member of the Home team suggested Wednesday that it needed just over a thousand rooms to gather meaningful data. But as soon as the speakers are getting to consumers in December, they’ll continue to gather this type of data, and improve over time.
Google also wants to get smarter at adopting the sound to particular situations. Google Home Max owners might crank up their favorite tunes when they get together with friends at night. But blasting the morning news briefing at the same volume might lead to a lot of spilled coffee. “Sound has to be contextual to your environment,” Chandra said.
This will also include raising the volume when there is a lot of ambient noise, and fine-tuning it based on the type of content the device is playing. “We want to build a contextual sound engine for you,” Chandra said. He also told Variety that smart sound won’t just be coming to the Google Home Max. Future Google Home iterations, and other devices made by Google, may benefit from this approach as well.
Google isn’t the only company looking to use technology to improve sound. Sonos first introduced a sound optimization technology called Trueplay two years ago. However, Trueplay requires users to walk around their room waving their phones while their speaker plays strange control sounds — not exactly a user-friendly approach. Apple also announced that its upcoming HomePod speaker will include automatic tuning to optimize for each and every room.
But truly smart sound, powered by machine learning, could in theory go even further. Startups like Doppler Labs have been introducing hearing aid-like products in recent years that are designed for more attentive hearing, and aim to help with the dinner party effect where you overhear snippets from the next table, but not what someone across from you is saying.
And then there are commercial speaker manufacturers like Meyer Sound, which has developed technology to transform rooms into sound stages appropriate to the music that’s playing at any given time. Thanks to a combination of speakers, microphones, and software, Meyer Sound’s systems can make a room sound like a church, a rock concert arena, or a small jazz club.
Chandra didn’t want to reveal too much about where Google is going with its own efforts to make sound smart, but he did say that the company is serious about the field: “Smart sound is going to be a long-term investment for us.”