That you can hold a phone up to music these days and identify it is amazing. But which smartphone song-finding app is more amazing? We researched and road-tested the two most popular song ID apps to see which was a better install.
Both Shazam and SoundHound offer free song ID apps for Android and iPhone, with Shazam covering other platforms, too. Shazam’s free iPhone version is limited to five song IDs, or “tags”, per month, while SoundHound is an ad-supported but otherwise free service. Both show artist and song information when they can identify a song, with purchase and YouTube links and bookmark options. SoundHound has a bit more visual build-out, but Shazam has the simple, clean look down. Other than that, they both serve the same purpose: to name the songs you’re currently hearing.
How Do These Apps Work?
In the case of Shazam, the exact detailed method for its song identification system is a trade secret. But the Shazam team has revealed some aspects of the system in white papers, and some have explained the basic algorithm.
Put plainly, Shazam contains a vast database of songs. Those songs are taken apart and analysed for their acoustic properties — specifically the “peak frequencies” at various points. When you capture 10 seconds of audio and submit it to Shazam’s severs, you’re also grabbing frequencies: notes from instruments, unique harmonies and the like. Other people may be talking, your car’s heating or A/C may be running, but if those notes and tones can come through, they should (in theory) still match up with certain combinations of frequencies found in Shazam’s database.
SoundHound’s explanation is a bit more vague, and the backing technology, Sound2Sound, also allows for humming or singing tunes for identification through Query by humming (QbH). The official about page suggests a system likely similar in outline to Shazam’s, but with perhaps different unique features being searched out.
“S2S performs recognition by extracting features from the input signal and converting them to a compact and flexible Crystal representation. This Input Crystal is then matched against a database of Target Crystals which have been derived from searchable content.”
Get your singing, humming or tapping chops ready, and you can try out QbH at Musipedia. In the meantime, let’s give Shazam and SoundHound some real-world tests.
Test 1: Driving, Heat Blasting, Android Held Over Steering Wheel
The realest of real-world app use we can imagine is actually a time when, all things considered, it would be better not to be using either one of these apps. So, by all means, hand over your phone to a passenger, or pull over to give your phone a few seconds alone with that song you’re digging.
But if you must click the big button on Shazam or SoundHound, and hold it just under the steering wheel while you drive, here’s how they would perform. For these tests, I loaded a CD, while driving with the heat set to its highest fan level, and activated each app at the same point in each song.
Queens of the Stone Age: “No One Knows”
Shazam: Nailed on the first try.
SoundHound: Wrong pick on the first try — “Ghost of Love” by Rasmus. Second try successful.[imgclear]
The Beatles: “You Never Give Me Your Money”
Shazam: Data/upload errors halted app three times; correct on fourth try.
SoundHound: Right on first try.[imgclear]
Jimmy Smith: “After Hours”
Shazam: Could not identify in three tries.
SoundHound: Right on first try.[imgclear]
Test 2: iMac Speakers, 1-1.5m Away from an Android
In a small, small office, one worker controls the Pandora mix for everyone’s soundtrack. I used my Nexus One to try and tag a few songs, then verified them with the station creator. People may have been quietly talking during some attempts, but not close to either the phone or the speakers.
Justice: “Newjack”
Shazam: Third try.
SoundHound: First try.
Passion Pit: “Sleepyhead”
Shazam: First try.
SoundHound: First try.
Saosin: “Come Close”
Shazam: Second try.
SoundHound: First try.[imgclear]
Test 3: iPod Speakers, iPhone Held 3m Away, Holding a Conversation
A very willing friend kept an awkward patter going, while I started at 3m, than 1.5m, then putting her iPhone (3G) right next to the speaker, to see how close I’d need to be if people were actively talking during a song.
The Social Network Soundtrack, “Hand Covers Bruise”
Shazam: Picked up when next to speakers.
SoundHound: Never got it.
Sufjan Stevens, “Too Much (Short Version)”
Shazam: Never got it.
SoundHound: Picked up next to speakers.
Big Star, “Back of a Car”
Shazam: Second try (about 1.5m).
SoundHound: Second try (about 1.5m).
What Did We Learn?
When it comes to the ideal conditions for identifying a song, it would seem a few rules stand out. For one thing, get your phone as close to the music source as possible. On another point, noise is better dealt with than conversation, so, well, shut your trap if you’d really like to know the artist and track.
I’ve used both apps outside these “official” tests, too, and I’ve come to think of Shazam as better with “popular” music, whereas SoundHound seems to have deep roots in jazz and other older, wider catalogues. That said, neither app was all that great with tracks that featured distortion, ambience and electronic effects — the more clear the melody, and crisp the singer’s voice, the better shot the app seems to have at it, which jibes with the underlying frequency-charting technology.
Given one choice, I’d go with SoundHound, as it seems dedicated to being free to use in the long term, glitched out less (though perhaps this is an Android-specific problem for Shazam), and picked up more songs that Shazam had no idea about. Shazam, on the other hand, seemed like a great app for anyone with tastes running toward pop music. I thought I’d read somewhere about Shazam’s technology used as a part of YouTube’s copyrighted material identification program, but I can’t find a relevant link at the moment.
Done your own side-by-side tests with Shazam, SoundHound or other song ID clients? Tell us which one you’d use in your car or favourite loud spot in the comments.
Leave a Reply
You must be logged in to post a comment.