Do Shazam And SoundHound Recognize What Radio Didn’t? posted on March 19th, 2014
By Sean Ross
It’s been a big year for Shazam.
The music recognition app is now a regular part of record label trade advertising.
At Country Radio Seminar in February, it was key in several attempts to apply “Moneyball”-type analytics to potential hit records. One of them was a formula by Stone Door Media Lab’s Jeff Green that eight or more tags-per-spin in a song’s second chart week could predict a top 10 hit.
One sees Shazam (or SoundHound or MusicID) being used in public on a regular basis now, to the point where it’s becoming a shared experience. Over the holidays, it was an app that helped me figure out that ELO’s “Long Black Road” was playing in a key scene in “American Hustle.” As we filed out, I spotted somebody tagging the same ELO song during the final credits.
I occasionally run into songs that stump all three of my music recognition apps. For the most part, however, they’ve sharply reduced the trouble that I used to go through to find out what song I was hearing. Some of those song quests ventured to the outer limits of music geekery. The most extreme involved buttonholing a Swedish consultant in the halls of the NAB radio show and humming a song I’d heard a decade earlier. (It worked.)
But mostly I did what regular listeners would have, if they’d been particularly taken with a song. I spent a lot of time waiting for somebody to pick up a radio station request line, back when there was still a person to answer them. Or I would negotiate with put-upon sounding station receptionists, who could never be bothered to hunt down the info themselves, and some who wouldn’t even transfer the call to the music director.
I always wondered if program and music directors knew just how careless their staffs were about flagging curiosity calls. To a PD, curiosity calls were always the first sign of passion on a new song. Even a few could be enough to save a song that was otherwise on the cusp of playability. Yet stations did so little to capture that information.
The willingness to make a phone call counted for a lot. Many listeners cheerfully go through their busy lives with only a vague idea of what artist actually sings that song they sort of like. Even among those who wouldn’t, however, complaints about stations that didn’t identify songs were a listener research staple for years. And the contradiction with also wanting less talk didn’t make them listeners any less sincere in their frustration.
Some broadcasters have become more cognizant of identifying songs in the last decade, especially the handful now using pre-recorded “song tags” for every song. But the irony is that radio has, for the most part, handed over both a research and a listener bonding opportunity. The apps have a broader reach than radio — listeners who would never call a radio station are Shazaming obscure recent ELO songs that they hear somewhere else anyway. But the information that we could have gotten from listeners by answering the phone is now being jobbed out, at a cost, to a third party as well.
Record labels have been consistently unsparing over the years toward radio’s lack of front- and back-sells. I have understood, in ways that labels did not always, why identifying every song wasn’t always good programming. But now Shazam, which was also on the dais at CRS, will “write new business” for doing the thing that labels wanted from radio.
This discussion led my Edison Research colleague Larry Rosin to suggest that radio station apps should include music recognition software as well. Beyond that, radio still has a lot of opportunities on a daily basis to tell listeners what they’re hearing, and to make them excited about it. Not relinquishing that franchise all together is part and parcel of protecting bigger franchises — music discovery and music creation.
If radio does not want to entirely abdicate the job of telling listeners what they’re hearing, they can do the following:
Make sure the “now playing” information on the Website and the station player actually works. And that the player and website refresh when there are new songs. And that weekend feature and other special programming songs are loaded in the system so that they display as well.
When we remind listeners that all of our song history is available on the station website, do it in conjunction with a real backsell, so that it doesn’t seem like just another cynical bid for web traffic.
Tell the morning show to actually acknowledge and interact the music, instead of having no seeming relationship to it.
Make station metadata robust and accurate. Our streaming players and websites often make the same gaffes as rookie DJs (attributing a major hit from a famous album to a greatest hits compilation, for instance). Song ID apps aren’t perfect; a friend recently found a Dean Martin song wrongly credited to “the Cat Pack.” But advocates of the FM-on-mobile NextRadio app are correct when they say that listeners expect more and better metadata than they’re getting from radio.
Radio hasn’t just provided music to listeners over the years. It has provided music expertise. Listeners now have more ability to find new music and find out more about that music. In that regard, fewer opportunities to be their hero exist. Those that still come radio’s way should not be lost.