SoundHound vs. Shazam, (sort of) scientifically tested

[captpackrat]

I installed the latest versions of Shazam Encore and SoundHound Infinity on a Motorola Xoom.  I then selected 50 songs with varying degrees of obscurity, played them back one at a time and asked each app to identify it.  I tried to use the same segment of the song for each app, usually the very beginning.  If it couldn't identify the song from the beginning, I fast-forwarded to the chorus and tried from there.

Shazam properly identified 31 of the songs, claimed it could not identify 19 and made no incorrect guesses.
SoundHound properly identified 32 of the songs, claimed it could not identify 10, and made 8 incorrect guesses.

SoundHound was much faster when it was confident about a song, but slower if it wasn't sure.  SoundHound was able to identify the most obscure stuff more often than Shazam, but it also made a number of mistakes while Shazam made no errors.  SoundHound crashed once during the test; Shazam performed flawlessly.


For those who want more information about the songs used and the specific results I present this handy chart:

Song Title	Artist	Shazam?	Soundhound?
Doctor Who	BBC Radiophonic Workshop	Yes	WRONG
Improperium	Bendictine Monks of Santo Domindo De Silos	Yes	WRONG
I See a Boat On The River	Boney M.	Yes	Yes
Metric Conversion	Bruce Haack	No	No
It Came Out Of The Sky	Creedence Clearwater Revival	Yes	Yes
Corngrinding Songs	Delphine Tsinajinnie	No	WRONG
Banzai Washout	Dick Dale & His Del-Tones	Yes	Yes
America Sings	Disneyland soundtrack	No	No
Cantina Band	The Evil Genius Orchestra	Yes	WRONG
Dracula's Deuce	The Ghouls	No	Yes
Everybody's Rich But Us	Glenn Yarbrough	No	No
White Sandy Beach of Hawai'i	Isreal Kamakawiwo'ole	Yes	Yes
Batman	Jan & Dean	Yes	Yes
Lily of the Valley	The Limeliters	No	No
Nitrous Burn Out	Man or Astro-Man?	No	Yes
Her Lazer Light Eyes	Nullsleep	No	WRONG
No Rewind	The Orchestra	No	No
Flash To The Rescue	Queen	Yes	Yes
August October	Robin Gibb	No	Yes
Snoopy For President	The Royal Guardsmen	Yes	Yes
People Mover	Servotron	Yes	Yes
Brazillian Love Song	Steel Drums of Trinidad	No	No
Am I In Sync	Steve Taylor	Yes	Yes
Toast And Marmalade For Tea	Tin Tin	Yes	Yes
Poisoning Pigeons In The Park	Tom Lehrer	No	Yes
Love And Fury	The Tornados	Yes	Yes
Livin' In The Sunlight, Lovin' In The Moonlight	Tiny Tim	Yes	Yes
Let's Pretend We're Bunny Rabbits	Magnetic Fields	Yes	Yes
Squirrels	The Beastly Boys	No	WRONG
The Little Green Men	George Morgan	Yes	No
America, F**k Yeah!	Team America	Yes	Yes
Say Man	Bo Diddley	Yes	Yes
The Blob	The Five Blobs	Yes	Yes
The Raccoon Song	Jan Hobson & Her Bad Review	No	No
The Bertha Butt Boogie	Jimmy Castor Bunch	Yes	Yes
Black Superman (Muhammed Ali)	Johnny Wakelin	Yes	Yes
Beans In My Ears	Serendipity Singers	No	Yes
Martian Hop	The Ran-Dells	Yes	Yes
In The Franches Mountains	Swiss Mountain Musicians	No	No
Syncopated Clock	Leroy Anderson	No	No
Get Up, Action	Digital Emotion	Yes	Yes
Does Your Chewing Gum Lose Its Flavor (On the Bedpost Overnight)	Lonnie Donegan	Yes	Yes
Purple People Eater	Sheb Wooley	Yes	Yes
Star Trekkin'	The Firm	Yes	Yes
Der Fuehrer's Face	Spike Jones	Yes	WRONG
Mr. Custer	Larry Verne	Yes	Yes
Dungeons and Dragons	The Dead Alewives	Yes	Yes
Jingle Bells	Singing Dogs	No	Yes
Banned From Argo	Leslie Fish	No	Yes
A Nightmare On My Street	DJ Jazzy Jeff & the Fresh Prince	Yes	WRONG

Comments

[nipper.livejournal.com]

So, what was your conclusion? Which would you recommend and which do you use?

[captpackrat.livejournal.com]

They were close enough it would be hard for me to make a recommendation. If you can tolerate mistakes, SoundHound seemed to a better job at the really obscure stuff and most of the time it was several seconds faster. SoundHound also claims you can hum or sing a song to identify it, something I didn't test. And it handles tablets slightly better (it can run in portrait or landscape, Shazam is locked to portrait).

On the other hand, Shazam is about half the size, important if your storage is limited.