Artificial intelligence applies sound thinking to audio content
Innovations poised to disrupt industry, open up user experience to amateurs
The Jacques Cartier Bridge lighting ceremony in May set a milestone in more ways than one. As the crowd celebrated Montreal’s 375th anniversary, more than 22,000 synchronized cellphones provided the sound system for the music.
That moment garnered a world record for the number of devices outputting sound in unison. The city can thank a homegrown AI-enhanced mobile app called AmpMe for pumping up the volume.
AmpMe was created by highprofile serial entrepreneur and Montreal-based Dragons’ Den (Dans L’Oeil du Dragon) denizen Martin-Luc Archambault. He launched AmpMe in 2015 as a “social music player like Sonos without the speakers,” he says. Since then there have been more than four million downloads and counting.
The AI component comes into play by adding watermarks inside music streams so devices can recognize and synchronize with each other to the millisecond, he explains. The premise is simple: the more devices that are connected, the louder and richer the output.
Archambault believes that hardware such as Bluetooth speakers may become as obsolete as digital cameras for the everyday consumer. “Most people take photos on their phones. In five years, everyone will also be carrying a speaker in their pocket. I think that if music can be played and shared in every centre of the world, it would do good things for humanity.”
The world of sound is getting a boost from AI in many ways. Montreal is one community in Canada that is rapidly becoming a centre for sound innovation, from 3D audio software to machinelearning-based post-production platforms.
Matthew Boerum is co-founder and CEO of Audible Reality, Inc., a developer of a 3D software solution that creates an immersive experience for anyone using headphones. With Audible Reality, users can experience another person’s voice or music as if they were in the same room.
When Boerum worked with virtual reality companies in the past, he realized they didn’t seem to understand the concept that in addition to video, sound had to be 3D. From that came the inspiration for Audible Reality. “Because it’s a software play, any developer or service provider can build the solution into their platform to replicate real-word hearing.”
It is only recently that the right technology has come into play to deliver 3D audio, he says. One that’s familiar to all is surround sound, which stops short of being truly immersive. As Boerum says, surround sound processes multiple channels of audio and code digitally over cables. “That technology is going out of style, because it’s no longer necessary to encode and decode over multiple channels. Now you can use a single digital stream.”
Boerum believes the biggest potential for their offering is in communications and media streaming. The AI component is an important differentiator. “Our solution uses AI to look at content and a person’s use to make an educated guess as to what they might want to listen to where. It’s mainly positioned to help commercial entities, such as music streaming, enhance the user experience.”
Another audio innovation poised to disrupt the sound industry is LANDR Audio Inc. In this case it’s post-production engineering. LANDR (short for left and right) is a machine-learning-based audiomastering platform that can be accessed for a low monthly fee that allows even the most amateur of musicians to sound engineer their creations to a professional level. The service has 1.2 million users in 190 countries.
“We designed it to relieve aspiring artists who didn’t have much money to produce their music,” says CEO Pascal Pilon. “Post production usually takes too long and is expensive. With LANDR, users can get immediate results at a cost that’s five to 10 per cent of what they typically would have to pay.”
The secret to its success is the machine-learning component that can analyze the way human engineers operate on songs and, based on the music’s features, automatically engineer them to a level that fits the context in which they will be heard. To date the platform has processed five million songs, and can publish content to 200 streaming services worldwide.
“AI is taking an industry dominated by sound studios to the point where a teenager learning to record can mix a song professionally without the economic hurdles,” he says. “You just have to play or compose your music and LANDR does the rest.”
Whatever the technology, it’s the art behind the sound that matters most to these dedicated audiophiles.
“For a while the art went away,” Boerum says. “Today opportunities are so cheap everyone can do it.”