Now There’s A Speaker That Listens Back–And Learns

More than a speaker, Cone’s most interesting feature is the way it learns to play what you want to hear.

Now There’s A Speaker That Listens Back–And Learns
[Image: Flickr user phil-jackson]

Even the best speakers are pretty dumb. They’re good at projecting sound, but that’s about it. Cone is about to change that. The newly announced smart speaker from Skype cofounder Janus Friis‘s new startup Aether learns from you as you use it. And it just might change the way people listen.


“Cone learns from everything you do, says Duncan Lamb, cofounder and chief product officer. ”If you turn up the volume or ask for a certain song, that means you probably like that song. Skipping a song means you probably don’t care to listen to it at the time.”

Lamb continues, “All these things are noted against time, day, and place in the house, which creates context for what people like to hear and when. Down the road, there are many more data points that we will start to look at factoring in, all of which will help Cone decide what the perfect sound for now is.”

In other words, if you listen to music loudly in the living room after you get home from work, Cone will adjust accordingly. If you move it poolside on the weekends, it knows that and plays music appropriate to the situation.

Cone is the first attempt by Aether to bridge the gap between consumption and the physical devices used to access the content that’s being consumed.

“Janus and I had both seen the IBM Watson Jeopardy demo and were trying to figure out how it could be that on the one hand, we have all this computing power with content, data, and context at our disposal, and on the other hand, we still have to explicitly tell our devices what we want them to do,” says Lamb.

Cone is extremely promising, but isn’t a slam dunk yet. The first hindrance is the speaker’s $399 price. Though competitive if it can deliver a deep and rich sound, it still isn’t cost effective for a lot of people. The second challenge is one of usefulness.


The latest trend in digital music discovery is mood-based listening–playlists tailored to a certain situation or event like “summer at the beach,” rather than a genre. Beats and Songza both play up this angle of curating based on a user’s location, reminiscent to what Cone is attempting, but without hardware and for a lot cheaper.

So, if Cone can perform flawlessly it will have lots of supporters singing its praise, but if the learning algorithms fail too many times, other apps and services like those mentioned will most likely end up filling people’s needs, even if they require more effort configuring.

Even though Cone is decidedly complicated on the inside, full of learning algorithms, its existence can be summed up with one word: Simple.

“We could both see the direction things were heading, which was away from computers and toward a personal collection of devices like phones, tablets, wristbands, weighing scales, thermostats,” says Lamb. “It was a really exciting promise, but the reality being a ton of housework, setting things up, syncing them, charging them, provisioning them with an increasing number of accounts. That time could have been better spent.”

Not having to think about what content to listen to could change the way we think about our existing music libraries and content sources. If the speaker really can deliver accurate results, however, it could sway whether someone pushes play instead of turning on a video game or other form of media.