4 ms·

This is the feature that is where it all started[1] > If you don’t turn it on, we won’t use your microphone to try and match TV or music when you write a statu

This is the feature that is where it all started[1] > If you don’t turn it on, we won’t use your microphone to try and match TV or music when you write a status update. If you do choose to turn it on and later decide it’s not for you, you can easily turn it off at any time. > Fact: Nope, no matter how interesting your conversation, this feature does not store sound or recordings. Facebook isn’t listening to or storing your conversations. > Here’s how it works: if you choose to turn the feature on, when you write a status update, the app converts any sound into an audio fingerprint on your phone. This fingerprint is sent to our servers to try and match it against our database of audio and TV fingerprints. By design, we do not store fingerprints from your device for any amount of time. And in any event, the fingerprints can’t be reversed into the original audio because they don’t contain enough information. The app does fingerprint the audio and send it to their server to match a movie or audio. But they say they don't save it. [1]: http://newsroom.fb.com/news/2014/05/a-new-optional-way-to-share-and-discover-music-tv-and-movies/ http://newsroom.fb.com/news/2014/05/a-new-optional-way-to-sh...

foxfired 10y ago
Quick note, now that I reread what I quoted, I noticed how they say they do not store sound or recordings. But what if it is converted to text? Notice below they say that the fingerprint cannot be reversed into the original audio because it doesn't contain enough information. Text cannot be converted to the ORIGINAL audio but it can be converted to audio. This may be just me being paranoid, but I do know that legalese is designed exactly to say something without saying everything. Anyway.
fencepost 10y ago
Huh, that would raise the interesting question of whether it's also able to "fingerprint" individual spoken words if they're distinctive enough. Admittedly this is probablymaybeperhaps only happening while you're writing a status update, but even so it's not that unlikely that some words/brands could be recognizable. After all, low-impact recognition of specific phrases is the core of a variety of things - "OK Google," "Hey Siri," "Hey Cortana"

foxfired 10y ago

Quick note, now that I reread what I quoted, I noticed how they say they do not store sound or recordings. But what if it is converted to text? Notice below they say that the fingerprint cannot be reversed into the original audio because it doesn't contain enough information. Text cannot be converted to the ORIGINAL audio but it can be converted to audio. This may be just me being paranoid, but I do know that legalese is designed exactly to say something without saying everything. Anyway.

fencepost 10y ago

Huh, that would raise the interesting question of whether it's also able to "fingerprint" individual spoken words if they're distinctive enough. Admittedly this is probablymaybeperhaps only happening while you're writing a status update, but even so it's not that unlikely that some words/brands could be recognizable. After all, low-impact recognition of specific phrases is the core of a variety of things - "OK Google," "Hey Siri," "Hey Cortana"