Hacker News Reader

Comment by xirdstl | Hacker News Reader

The first thing the article says is that users are probably aware Google stores all of these recordings. I, for one, was not aware. Am I really in the minority?

View on HN

I was surprised and dismayed to discover that Skype does something similar.

Skype stores your voice mails and video messages forever[1]. This is something that they started doing 2-3 years ago and few people seem to be aware of it. Unlike Google, they don't provide any way to delete them.

[1] Details: Clicking on Preferences -> Privacy -> Delete history (OS X) or Options -> Privacy Settings -> Clear history (Windows) pretends to delete the voice/video messages but it merely hides them from your view. If you re-install Skype on the same computer or run Skype on a different computer, all those "deleted" voice mails and video messages re-appear.

That is not good. The least they could do is be explicit and clarify you're deleting the local copy (which I hope is what they're doing).

Google, Siri, Cortana and Amazon Echo all store your voice recordings. They are explicit about this in the fine print. Google and Amazon allow you to listen to these recordings yourself[1][2]. Not sure about Siri or Cortana.

[1]https://history.google.com/history/audio [2]https://www.amazon.com/gp/help/customer/display.html?nodeId=...

What's more, you may get to listen to the recordings, but you're not the only one: https://news.ycombinator.com/item?id=9101875

Siri as of 2013: http://www.wired.com/2013/04/siri-two-years/ "Whenever you speak into Apple’s voice activated personal digital assistant, it ships it off to Apple’s data farm for analysis. Apple generates a random numbers to represent the user and it associates the voice files with that number. This number — not your Apple user ID or email address — represents you as far as Siri’s back-end voice analysis system is concerned.

Once the voice recording is six months old, Apple “disassociates” your user number from the clip, deleting the number from the voice file. But it keeps these disassociated files for up to 18 more months for testing and product improvement purposes.

“Apple may keep anonymized Siri data for up to two years,” Muller says “If a user turns Siri off, both identifiers are deleted immediately along with any associated data.”"

Google asks you if you want to store your data to improve their model. Presumably they augment their model of you with the voice data and corrections you supply (voice-typed words are blue in android, and selecting one brings up a dropdown of alternate words that would be stupid not to record...).

I guess I always knew this data was being recorded, but honestly I would rather that Google not have released it via this interface. My Google account is not as well-secured as Google's backend, and if someone compromises my account I don't want them to be able to download recordings of my voice saying things (I probably have enough google searches on voice to make a soundboard for anything).

But, I can also appreciate that few users will appreciate the privacy implications of choosing to personalize the google voice model without a service like this, so...

> I guess I always knew this data was being recorded, but honestly I would rather that Google not have released it via this interface. My Google account is not as well-secured as Google's backend, and if someone compromises my account I don't want them to be able to download recordings of my voice saying things (I probably have enough google searches on voice to make a soundboard for anything).

I would argue, though, that the majority of people want to keep control of their data as much as possible. The responsibility to take care of it and it's security is just a corollary of that.

>I would argue, though, that the majority of people want to keep control of their data as much as possible

This seems like a nonsensical statement in 2015. What basis do you have for it? Most people hand over control of their data without a care in the world.

> This seems like a nonsensical statement in 2015. What basis do you have for it? Most people hand over control of their data without a care in the world.

Have you watched the news lately? At least in Europe that is most definitely what users want and what the laws are giving them the right to and the courts are enabling them to do.

Unless of course, you are trying to derail this argument by meaning whenever something isn't saved on a hard drive in my safe it's out of my control. Which would be complete BS and irrelevant to the discussion (which is about "is Google trying to give control over their data to the users").

Fyi,

Skype, Google, YIM, AIM, etc. all store large portions of your communications far longer than you realize and their "clear" mechanisms frequently are local-only. [i.e. Skype, reinstall, things reappear]

> The first thing the article says is that users are probably aware Google stores all of these recordings. I, for one, was not aware. Am I really in the minority?

Nah. Most people are convinced the deletion stuff actually works and/or people wouldn't engage in this sort of behavior and are surprised when they find out.

The reason people don't care about privacy is they don't really comprehend what they've lost and how extensive it is.

I was not aware of this until a couple days ago, when I stumbled upon that (OP) article. Surprisingly, though, my privacy settings were already configured to not keep these recordings.

I was not aware of Google's policy for this issue, but I knew that Apple keeps all recordings for at least 2 years... so I was thinking that they did about the same.

Last I read, that was anonymized data. Or have things changed?

Anonymized voice recordings? I can't recall if the last article that I read about that a while ago mentioned it, but either way, I would say that it is pretty much impossible to completely anonymize a voice recording.

Nor would anonymising a voice recording (presumably by doing some sort of frequency-domain transform) be of much use to them for the sort of thing they're using them for - training speech recognition on a wide variety of different voices.

It's virtually impossible to anonymize voice. Except by doing voice to text, and then synthesize voice.

Even if not "completely", there is a difference to me as a consumer

The data has a common ID, but that ID isn't linked to your Apple ID, and you can change the ID at any time.

Of course not. Most people don't even think about this stuff.

Well, I don't know. A typical HN reader should not be surprised by this. You know technology, you know how Google operates, it should be fairly obvious that they store everything, forever.

When I saw the article, my reaction was "hey, thanks for letting me know what the link actually is". I was not surprised at all by it. I sort of expected it.

I assumed they did something to improve matching, but I never imagined they stored the 'raw' (as in actual voice, albeit probably compressed) recordings.

So they store Google hangout video chats forever?

You might know something when thinking about it but never have been aware of it. "Oh yeah, I guess all this time they must have been doing that."

If you've kept up with internet privacy news the last few years from things like the snowden or ashley madison scandals, how is it not your assumption that everything that is exposed to the internet is stored and indexed somewhere forever?

There's a psychological difference between knowing it in the abstract and actually being confronted with and able to play back that history.

And least for this, there's a way to delete it.

How do we know for "sure" that it is deleted?

There are laws against retaining user-data that the user has explicitly requested to be deleted and Google spends a lot of engineering time making sure, that is really the case (which is non-trivial e.g. with failing drives that you can't access anymore. You still have to delete the data from them, meaning you must know, at every point in time, the physical drive that a users data is saved on). I think 30 days is the guarantee they make.

as per my previous examples, with the ashley maddison website users would pay to get their information "deleted" and it ended up not being deleted at all, just hidden.

Perhaps that's a fair point if you are the kind of user that keeps up with all that you've detailed. To say that users who use of Google's voice controlled features are probably aware of it is another thing entirely, and highly unlikely.

I'm a bit surprised myself, more at the potential storage cost than the creepiness. Even at a low bit-rate, that's 10 MB/hr, which would add up quickly if many people are actually using Google voice search.

Google says Google was doing a bit over 3 billion queries a day. Let's say 1% are voice (which is prolly way high). 30M voice queries at 5 seconds each is 42K hours. At 10MB per hour that's 420GB per day. Seems rather trivial.

Especially compared to YouTube, which was estimated to be growing at roughly 200TB per day 3 years ago:

https://sumanrs.wordpress.com/2012/04/14/youtube-yearly-cost...

every time I use voice search on my mobile (the only place I use it) I get a toast popup that says 'saving audio to $email_address'. So I knew about this.

I use an android phone, I have however noticed that voice search on the desktop doesn't issue this warning.

Reply on news.ycombinator.com