Among the technical tools apparently being used by iHeartMedia in its dramatic organizational restructure is a music-mixing A.I. system built by Super Hi-Fi.
It’s being reported on today by the Washington Post and was described in some detail in an earlier Radio World interview with iHeart’s chief product officer. It’s not clear the extent to which Super Hi-Fi is at the core of iHeart’s AI, given that company officials are declining to talk about such specifics, but it seems likely to be a central component given that Super Hi-Fi was being tried out in its streaming platform.
The Post reports: “The system can transition in real time between songs by layering in music, sound effects, voice-over snippets and ads, delivering the style of smooth, seamless playback that has long been the human DJ’s trade. The Los Angeles-based Super Hi-Fi, whose clients also include the streaming fitness service Peloton, says its ‘computational music presentation’ AI can help erase the seconds-long gaps between songs that can lead to ‘a loss of energy, lack of continuity and disquieting sterility’.”
The Post described patents that it says “reduce the art of mixing music to a diagram of algorithmic tasks,” including a system called MagicStitch that assesses songs’ technical characteristics, blends songs and interjects other elements. The reporter describes a demo given by the company and points out a comment by iHeart’s chief product officer, Chris Williams, in an interview by Radio World that “virtual DJs” that could seamlessly interweave chatter, music and ads were “absolutely” coming, and “something we are always thinking about.”
This caused us to take a fresh look at that 2018 Radio World interview.
Super Hi-Fi describes itself as a company “dedicated to the creation of powerful artificial intelligence tools to help digital music services deliver amazing listening experiences.” In the earlier story, Williams described how Super Hi-Fi would add “perfect transitions,” “sonic leveling” and “gapless playback.”
Williams described the technology at the time as applied to iHeartRadio streams rather than over-the-air broadcasts, but the conversation presaged the impact on the latter.
“We’re eliminating the periods of silence that users currently experience within streaming music to create an experience that mimics the polished production of live radio,” Williams told RW at the time. “We’ve audited the user experience across all the major services and the average gap is 4-6 seconds between the end of one song and the start of another.”
He said the perception of the gap can be even longer across songs with really long, quiet fades or silence at the end. “This new A.I. takes all this into consideration to create the perfect song transitions just as a seasoned radio programmer or DJ would do.”
The technology also levels the volume across songs from different decades, he said at the time.
“This is important because music plays a role in setting a mood and amplifying an experience. Silence between every song and jarring changes in volume breaks the spell and takes a user out of the flow of their experience. It’s an unwelcome disruption that we can eliminate so that the music does what we intend it to do — enhance the moment.”
Williams said in the 2018 interview that these are not cross fade or segue tones, traditional methods the industry uses to solve a transition problem. “Our solution considers every transition discretely, analyzing the song ending as well as the song playing next,” he told RW. “The transition point for a single song is going to vary depending on what track is following, it is dynamic for each unique transition. Our transitions factor in energy, tempo, instrumentation, vocals, processing, volume, production values and hundreds of other attributes for one transition on the fly.
“Each time a new song is ingested, the A.I. learns the characteristics of that track and how to best transition it with every other song in the library, similar to the masterful capabilities of our on-air programmers.”
Williams said there are two parts of programming that affect the user experience and have to be considered: curation and presentation. “The curation, or song selection, is still based on our custom algorithm, which is influenced by the curation expertise of our world-class radio programmers. The presentation, or how the songs are stitched together, is what’s being enhanced using the Super Hi-Fi A.I.”
But asked how “artificial intelligence” could be used in a radio operation, he replied: “For a streaming music service, it allows us to scale this concept across millions of songs and billions of unique transitions in a way that isn’t possible if it had to be done by hand,” Williams said. For radio, “We would have to resort to one of the static solutions versus the dynamic approach that we have adopted.”
Notably, Radio World asked Williams, “Do you envision a day when iHeartRadio streams will have virtual DJs, complete with Casey Kasem/Gary Owens voices delivering chatter, tidbits about the song/artist or even local weather before the song plays?”
He answered, “Absolutely. Being able to add in personality, branding, artist messages and weave them all together with the music in a way that is seamless and respects the music is something we are always thinking about.”
CENTERS OF EXCELLENCE
This week’s Washington Post article quotes the co-founder of Super Hi-Fi, Zack Zalon, saying its system won’t trigger massive job cuts and could lead to new opportunities, but also said he expects that, in a few years, computer-generated voices could read off news, serve interviews and introduce songs.
We note, too, that Super Hi-Fi was not mentioned by name in the recent iHeart announcement, which described “technology- and AI-enabled Centers of Excellence” that consolidate functional areas of expertise “in specific locations to deliver the highest quality products and services.” It did mention “the hundreds of millions of dollars in investment [iHeart] has made in building out the company’s core infrastructure, in addition to strategic technology and platform acquisitions like Jelli, RadioJar and Stuff Media.”
Meanwhile, iHeart spokeswoman Wendy Goldberg was quoted by the Post this week saying that its technical solutions allow the company to free up programming people for more creative pursuits, “embedding our radio stations into the communities and lives of our listeners better and deeper than they have been before.”