Automatic Content Recognition via Audio

After some discussion last night over dinner, I decided to spend an hour in front of 2 popular TV shows on a major network to see how the audio sync apps worked.  I ensured the conditions were the same for every app, testing each of them at least 3 times the same distance from the TV with the sound at a normal level and NO background noise.  I repeated the test again on a second show.  I tested them all on real-time performance on the West Coast (no DVR'd content).  Here's what I found:

- Yahoo IntoNow.  Recognized the show within 3-5 seconds every time.  No mistakes.  No delays.  Check-in capability only, but quick and accurate.

- TVplus.  Worked 5 of 6 times.  Recognized the show within 3-5 seconds.  No errors, 1 failed attempt.  Synchronized content experience.

- Umami.  Worked every time, recognizing the show within 3-5 seconds.  No errors.  Checkin capability only.

- ConnecTV.  Did not work even once.  For this app, I did try turning up the volume to no avail.

- Viggle.  Did not recognize the first major network show (0 for 3 attempts).  Did recognize the 2nd show 2 of 3 times.  Took a bit longer than the other apps (3-5 seconds of listening, 3-5 seconds to match).  Check-in capability only.


As an industry, if we are going to get consumers to adopt these kinds of experiences on a mainstream basis (ie crossing the chasm), the ability to recognize the show needs to be QUICK and ACCURATE.  I still think audio should be a last resort and the industry should focus on synchronizing with the set top box or the "OS" of the OTT delivery (Netflux, Hulu, etc).  Blu-ray is of course already easily and very accurately"sync-able".

Once we have accuracy and speed, we can drive better synchronized content use cases (vs. check-in only) for both live and recorded (or on-demand delivered) content.  Then we have a real shot at convincing consumers to pick up their 2nd screen and launch their favorite app while they watch TV.