
"Girls have got balls.
They're just a little higher up that's all."
--Joan Jett
They're just a little higher up that's all."
--Joan Jett
Musings by Pheme Ossa
Among the great joys of free, pirate and indie radio is hearing non-mainstream music. And among the frustrations is being unable to identify a song or artist to find more of their music. It's particularly difficult with instrumental music or vocals for which you can't hear the lyrics clearly enough to Google, or the lyrics are in another language.
Even some licensed radio stations are too damned hip to identify songs and artists. Despite complaints from listeners, some stations have adhered to the "no back announcing" policy that is just as stupid now as it was decades ago when some boneheaded consultants decided that somehow identifying songs already played would slow down the momentum, or would cut into advertising time, or would spoil the station's carefully cultivated hipper-than-thou mystique, or would require actual live human beings to identify songs.
There's probably a way to automate song IDs as well, using ID3 metadata to provide info for a computer voice to read. That'd be a great way to make Jack FM even less interesting.
Every few years the same story is recycled in some major news outlet.
"It just makes sense to do it (identify songs played)," said Dan Mason, the president and chief executive of CBS Radio. "(We) probably underestimated at the time how much people really wanted that information," he said of the no-clutter trend of the 1980s that eliminated most DJ breaks to identify songs.
--(The New York Times, CBS Radio Reminds D.J.'s to Identify Songs: 'When You Play It, Say It', May 29, 2011)
"When you work in a record store and someone comes in and starts humming something and you're supposed to guess what it is, that's frustrating," said Carl Rosenbaum, president of Flip Side Records, which operates 17 retail outlets in the Chicago area.
--(From the Los Angeles Times, 'What Was That Last Song?' : Record Industry, Radio Deejays at Odds Over Song IDs, April 11, 1989)
Opinions vary, natch:
"Don't know about you, but I get irritated when you hear presenters tell you what an obvious song was, for example "That was Meatloaf and I'll do Anything for Love" ... I remember John Myers once say if he ever heard a presenter say 'and that was The Beatles with Hey Jude" he'd fire them!"
--Some bloke on some other naff site.
With record stores mostly part of ancient history, to whom does one hum to name that tune?
How 'bout Tunatic, Midomi or Shazam?
(Update 12/10/11: Forgot to include the smart phone app SoundHound, about which Gizmodo sez "Unlike Shazam, it will hazard a guess at just about any tune thrown at it, whether it human or speaker-created, and it'll succeed most of the time.")
A couple of years ago Evil Elvis tipped us to Tunatic, a music recognition program that users reported worked surprisingly well, even for identifying songs off scratchy shortwave reception. Since then, Midomi, a comparable web-based music recognition utility, has appeared. We finally gave it a go recently.
Another recommended music recognition tool is Shazam, which appears to be available only as an app for mobile devices. Those of you who are less paranoid than we are may wish to try the Shazam or Midomi mobile versions for your personal entertainment/tracking devices.
The web-based version of Midomi uses a familiar Flash interface. It seems to assume users will hum or sing into their computer's microphone. If you record music off-air using a mic this would work, but not as well as a direct connection (see our experiences in the off-air recordings notes below). However you can easily set it to recognize music played from your own collection, or from YouTube or any other web source by setting your audio controls appropriately. For example, with a typical Windows PC, pull up the Recording Control and choose Mixer Balance (rather than the Line In option you'd use to record directly from your shortwave receiver). After Midomi suggests matches, many of the suggested songs are available in 30 second streams for immediate comparison and confirmation.
After a trial run with a few YouTube videos and mp3's off my computer, I began to be skeptical that Midomi was cheating and peeking at my mp3 metadata so I tried a few songs with no ID3 data, but it worked anyway. It appears to rely heavily on comparisons with an existing database of songs. That database already includes a large collection of familiar and eclectic music and it was surprisingly easy to identify music I'd never heard before.
First, the easy stuff:
- Pavarotti's "Nessun Dorma" = Score! But that's an easy one.
- Liz Phair "Fuck and Run" = Score!
- Black Eyed Peas "Pump It" = Fail on the first try. I'd have scored the original "Misirlou" as a winner, but "Chain Reaction" by Young Divas? Drop and gimme 20, Midomi.
- Second try for "Pump It" = Score! But I had to play it from 30 seconds into the song for Midomi to recognize it.
- Geto Boys "Damn It Feels Good To Be A Gangsta" = Score!
- Dead Kennedys "Too Drunk To Fuck" = Score!
- Nouvelle Vague version of "Too Drunk To Fuck" = Score!
- NON "Total War" = Fail. Close, but Midomi picked the somewhat similar "Phoenix" by NON. To be fair, the versions of those songs in Midomi's database aren't very good, mushy without much bass so the rat-a-tat drums in "Total War" were buried in a fog of white noise.
- "Do You Want Total Queer" = Fail. A little surprising it didn't default to the NON original for this obscure but brilliant spoof of "Total War".
- "Dirge" Death in Vegas = Score!
- "I Walk On Gilded Splinters" Dr. John = Score!
- "Dus Bahane (Karke Legaye Dil)" = Score! We dig the Hindi-pop and so does Midomi.
- "Do You Fear Sleep?" The Moscow Coup Attempt = Fail. Tried twice at different points in the song. Great song but probably too obscure so there's no copy in the Midomi database.
- "Radio Junk" Yellow Magic Orchestra, from Radio Junk programme #1 = Score!
- "Once In A Lifetime" cover by P.M. Dawn, also from Radio Junk #1 = Score!
- "Streets of Calcutta" Ananda Shankar, from Radio Junk programme #2 = Score!

Okay, enough easy stuff. Let's try some off-air recordings, including some songs I didn't recognize. Starting with the best signals and working downward toward the weaker, noisier signals and those for which my receiver wasn't tuned quite on frequency for sideband broadcasts.
- "Europa (Earth's Cry Heaven's Smile)" Santana = Score! From Radio Waves International Nov 13, 2011 show on 7415 with an excellent AM signal.
Wolverine Radio's Nov 13, 2011, train theme show, with strong signal peaks but fadey:
- "Midnight Train to Georgia" Gladys Knight & the Pips = Score! Pretty impressive for single sideband audio with static, fades and some distortion due to my noise reduction software, applied after recording to cut down the heavy fog of white noise.
- "Love Train" O'Jays = Score! Again, same Wolverine show, even worse static and fading.
From XFM off-air recordings:
- Tesla "Call It What You Want" = Score! From XFM July 4, 2011 show, fair to good conditions.
- Alice in Chains "Would" = Score! Same XFM show.
- "Valley Girl" Moon Unit = Score! XFM July 5, 2011, again, fair to good conditions.
- Right after "Valley Girl", Professional Murder Music "Slow" = Score! I'd never heard of this one before, couldn't even copy it well enough to Google the lyrics.
- "Too Drunk To Fuck" Dead Kennedys = Score! Not the most familiar version but Midomi nailed it.
- "The Red Weed" = Score! From unid relay of Jeff Wayne's "War of the Worlds" musical, Feb 2011. This was an excellent AM signal and audio.
- "I Love It When You Call Me Names" Joan Armatrading, from High Plains Relay Service, Dec 2011 = Score!
- "Cotton Eye Joe" Rednex = Score! From Feb 2011 Bust-a-Nut Radio show, with very muddy copy on my end.
- "You Keep Me Hanging On" Vanilla Fudge = Score! From Dec 2010 relay of Radio First Termer. Impressive because my recording was so weak I can't even ID the station broadcasting this relay.
- "Smoke! Smoke! Smoke! (That Cigarette)" = Fail. From unid station, Dec 2010, this was as near impossible as it gets, what I'd call an overall signal (on SIO or SINPO scale) of 2, just barely audible occasionally and able to copy only familiar songs. Only the "Smoke! Smoke! Smoke!" chorus was audible through static, and the pitch was off because the receiver was mistuned on a USB signal. This appears to be the practical limit for Midomi.
- "Antenna" Sonic Youth = Fail. From my own poor recording of our very own KBOX Radio Paranoia relay by WEAK, late 2010 or early 2011. Another overall 2 signal.
- Primitive Radio Gods "Standing Outside A Broken Phone Booth With Money In My Hand" and "Fading Out" = Fail. Very poor recording with mic up to speaker, barely audible.
- "Cruisin'" Mike Nesmith = Score, on second try! My recording, good signal strength but a lot of RFI. From Nov 26, 2011, High Plains Relay Service show just before relay of Gumby Radio. I didn't recognize this song myself. All I was able to copy was something about Lucy and Ramona, so Midomi helped.
- Gumby theme song = Fail. From my own recording of Gumby Radio, good copy but possibly not in Midomi database. No luck with my studio quality recording either.
- "Bend Me Shape Me" Hello = Score! From my Nov 26, 2011 recording of Gumby Radio.
- "Gumby Is Dead" Jeff Capo = Fail/Score? Failed on off-air recording. But scored, sort of, on studio recording. Midomi suggested The Sunbeat Revival - "When I Think Of You(I Smile Neon Light)". Sure enough, the intro is virtually identical. Jeff Capo's version came out first, so perhaps Sunbeat Revival sampled the intro. To be fair, this is a fairly obscure song.
- "Blockhead" Devo = Score!
- Fiona Apple live with audience sing-along of Gumby (Gunji) theme = Fail. No surprise, too obscure.
- "Bend Me Shape Me" by The Models = Score on second try! Failed on instrumental intro, but scored on opening lyrics. Very impressive. An obscure version, in nearly atonal singing style with mistuned audio from my receiver. (See: Girls in the Garage, Vol. 1)
- "Twist and Shout" Beatles = Score! Easy peasy.
- "Bend Me Shape Me" by the Rubinoos = Fail. Very similar to familiar and popular American Breed version, but no joy, including with studio recording.
- First up, I tried scatting a Muppet's style version of "Mahna Mahna". Midomi came close. Among the seven suggestions was a salsa styled "Mana Mana" by Ismael Rivera, which was pretty close. "Made In Japan" by Pato Fu? Not bad, it does indeed incorporate the "Mahna Mahna" bit. Gloria Estefan's "Conga"? Nah. And Phil Collins "You Can't Hurry Love?" Gimme a break. But let's blame my horrible humming. Let's call it score.
- Next, "Twinkle Twinkle Little Star" = Score! I wonder if it can do that alphabet song?
- First try at "Daisy, Daisy" = I'm sorry, Dave, it's going to go 100% failure. Midomi suggested "Iris" by Biagio Antonacci, "Yo Quisiera" by Reik, and "With You" by Chris Brown. Midomi doesnt' seem to like tuneless lower register voices.
- Second try at "Daisy, Daisy" = Score! Midomi seems to prefer the upper register.
- "Guantanamera" = Score! I actually sang "One ton tomato, oh he's a one ton tomato".
- Lady Gaga "Paparazzi" = Score! And I'm pretty sure Gaga would get a restraining order against my singing voice.
- "Blue Christmas" = Score! But Midomi suggested the Celine Dion version rather than the Elvis via Andy Kaufman version I thought I was singing.
- "Feliz Navidada" by Jose Feliciano = Score! It even recognized the ever popular "Fleas on my dog" variant.
Also, with almost every hummed, scatted and tortured sprechgesang version I tried, Midomi suggested around half a dozen possibilities. The one exception was "Feliz Navidad", which Midomi nailed without alternative suggestions. So if you go that route for an unknown song, plan on sorting through a few possibilities before finding your song.
Overall, definitely the hot tip for identifying those songs we hear on the funny bands, or even for you better-behaved SWLs who prefer to avoid the oh-so-melodramatic pirate frequencies and may want to send a detailed reception report to snag a QSL from Radio Havana Cuba, Radio Australia or other shortwave broadcasters featuring music programmes.