Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019) • Hacker News Board

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#101

Post by gruez » Mon, Jan 16, 2023, 9:56 PM UTC

brian_cloutier wrote: post #77

candiddevmike wrote: post #60

Earlier quoted context omitted.

How would you know when they do this? Buried inside a "We've made some updates to our Terms of Service" email?

There are a lot of people who run proxies like pi-hole while using these devices. It only takes one of those people to notice a sudden increase in traffic and post that finding somewhere. If you discover that Alexa is uploading information that it shouldn't be and blog about it you can be almost guaranteed to hit the front page of HN, these kinds of posts appear from time to time: https://news.ycombinator.com/item?id…

>It only takes one of those people to notice a sudden increase in traffic and post that finding somewhere.

Not if they perform speech speech to text on-device and send the parsed results (only a few kb). If you really want to keep things hidden you can perform all the recognition/inference on-device and only send the topic (eg. [1]), which is only a few bits of information.

[1] https://github.com/patcg-individual-drafts/topics/blob/main/...

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#102

Post by Rimintil » Mon, Jan 16, 2023, 10:03 PM UTC

googlryas wrote: post #79

itqwertz wrote: post #50

The real Alexa product is the wealth of data that the device can provide with 24 hr surveillance. The shaky voice recognition/command model is merely just the Trojan horse. Amazon now knows who is watching the Super Bowl and will treat you as a consumer they can advertise to. Alexa skill developers were sold a promise of platform promotion, quality, and customer interaction. The whole consumer experience for adding c…

Far more people have smart phones using one of 2 OSes, and unlike smart home devices, people generally take these with them wherever they go - to the store, to the bathroom, to sleep, to the doctors office, etc. Do you think cell phones are trojan horses too?

They certainly do make for easy tracking down to <1m. So they make a good target for surveillance states.

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#103

Post by gruez » Mon, Jan 16, 2023, 10:04 PM UTC

htag wrote: post #87

googlryas wrote: post #79

Earlier quoted context omitted.

Far more people have smart phones using one of 2 OSes, and unlike smart home devices, people generally take these with them wherever they go - to the store, to the bathroom, to sleep, to the doctors office, etc. Do you think cell phones are trojan horses too?

You don't? * How many apps ask for location data and background location data, when the use case for such permissions are questionable. * Look at how much FB's revenue decreased when it's consumer tracking efforts were given the smallest roadblock on iOS. * Law enforcement has access to a variety of tools that track individuals based on their cellphones. [0] * Google Maps requires an incredible amount of connectivity…

>* Look at how much FB's revenue decreased when it's consumer tracking efforts were given the smallest roadblock on iOS.

How much did it drop? By "smallest roadblock on iOS", I'm assuming you mean app tracking transparency, which was implemented in ios 14.5 (released April 26, 2021). According to this graph[1], facebook revenue for Q2 2021 was up from last quarter.

[1] https://www.statista.com/statistics/277963/facebooks-quarter...

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#104

Post by bastawhiz » Mon, Jan 16, 2023, 10:10 PM UTC

gruez wrote: post #101

brian_cloutier wrote: post #77

Earlier quoted context omitted.

There are a lot of people who run proxies like pi-hole while using these devices. It only takes one of those people to notice a sudden increase in traffic and post that finding somewhere. If you discover that Alexa is uploading information that it shouldn't be and blog about it you can be almost guaranteed to hit the front page of HN, these kinds of posts appear from time to time: https://news.ycombinator.com/item?id…

>It only takes one of those people to notice a sudden increase in traffic and post that finding somewhere. Not if they perform speech speech to text on-device and send the parsed results (only a few kb). If you really want to keep things hidden you can perform all the recognition/inference on-device and only send the topic (eg. [1]), which is only a few bits of information. [1] https://github.com/patcg-individual-dra…

While plausible on paper, it's not practical unless they jam an order of magnitude or two more compute into the devices. To get reasonable accuracy (i.e., enough to be able to use for profit) from any casual speech, the current models run far from realtime on a modern MacBook. You're not going to squeeze reasonable accuracy from the tiny processor on the devices in the world today, even if you record and process async as a way to hide from people inspecting traffic.

Edit: it's worth noting that this dramatically increases the cost of the device. They'd need to be able to see a way to recoup those costs if they eat the additional hardware cost. But that's silly for a company that's literally in the business of cloud computing and where the goal of the hardware is to hide what you're doing. When will people start asking why there's a full GPU in their Echo?

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#105

Post by incanus77 » Mon, Jan 16, 2023, 10:24 PM UTC

ars wrote: post #99

incanus77 wrote: post #92

Earlier quoted context omitted.

Per the Apple article, the detection is customized to the user who setup the device and went through an (on-device) enrollment training. > We compare the distances to the reference patterns created during enrollment with another threshold to decide whether the sound that triggered the detector is likely to be "Hey Siri" spoken by the enrolled user. > This process not only reduces the probability that "Hey Siri" spoke…

You seem to have a strange blind spot for Apple. Amazon is doing the exact same thing: Only sending to the cloud what it believes to be a wake word. The stuff you are writing is just not true. > This process not only reduces the probability that "Hey Siri" spoken by another person ... Contrast with Amazon Siri is personal on a single phone, while Alexa is meant to reply to anyone in the room. You need to figure out w…

It's true, I have a long Apple history and I do think that Apple means what it says about prioritizing privacy over convenience (and in many cases, accuracy — Siri is quite frustrating at times). They have taken a not-cloud-first approach which causes the quality to suffer for the tradeoff of gaining plausible deniability with storage of personal data and emphasis on user privacy.

But I still think Apple's approach is radically more aggressive on the "what it believes to be a wake word" angle. And the fact that wake words will never be validated in the cloud is arguably (I'm trying to make the point) better than sending any wake words to the cloud. Less data in the cloud is better for user privacy.

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#106

Post by googlryas » Mon, Jan 16, 2023, 10:57 PM UTC

htag wrote: post #87

googlryas wrote: post #79

Earlier quoted context omitted.

Far more people have smart phones using one of 2 OSes, and unlike smart home devices, people generally take these with them wherever they go - to the store, to the bathroom, to sleep, to the doctors office, etc. Do you think cell phones are trojan horses too?

You don't? * How many apps ask for location data and background location data, when the use case for such permissions are questionable. * Look at how much FB's revenue decreased when it's consumer tracking efforts were given the smallest roadblock on iOS. * Law enforcement has access to a variety of tools that track individuals based on their cellphones. [0] * Google Maps requires an incredible amount of connectivity…

I certainly think phones have more spying potential than Alexa/smart devices. But my point is very few people bring up "but what about spying?" whenever smartphones are discussed, with the implication that these devices should be avoided because of that.

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#107

Post by gruez » Mon, Jan 16, 2023, 11:03 PM UTC

bastawhiz wrote: post #104

gruez wrote: post #101

Earlier quoted context omitted.

>It only takes one of those people to notice a sudden increase in traffic and post that finding somewhere. Not if they perform speech speech to text on-device and send the parsed results (only a few kb). If you really want to keep things hidden you can perform all the recognition/inference on-device and only send the topic (eg. [1]), which is only a few bits of information. [1] https://github.com/patcg-individual-dra…

While plausible on paper, it's not practical unless they jam an order of magnitude or two more compute into the devices. To get reasonable accuracy (i.e., enough to be able to use for profit) from any casual speech, the current models run far from realtime on a modern MacBook. You're not going to squeeze reasonable accuracy from the tiny processor on the devices in the world today, even if you record and process asyn…

> While plausible on paper, it's not practical unless they jam an order of magnitude or two more compute into the devices. To get reasonable accuracy (i.e., enough to be able to use for profit) from any casual speech, the current models run far from realtime on a modern MacBook. You're not going to squeeze reasonable accuracy from the tiny processor on the devices in the world today, even if you record and process async as a way to hide from people inspecting traffic.

Do you really need 100% accuracy here? This isn't like cops setting up a wiretap. Google isn't waiting for you to slip up and admit that you like funko pops or whatever. If you're constantly talking about your cat, or wanting to get a car, that's all they need to target ads to you.

Also, the processing doesn't have to be real time. It doesn't matter that google learns about your cat 8 hours late because the device is running its ML models in the background while you're asleep. If the device picks up 3 hours of speech per day, it only needs to process at 1/8x speed to catch up. On the off chance you have a house party and it's picking up 6 hours of speech, it can always buffer it for later, or drop it altogether (see above paragraph about how it doesn't need to pick up everything).

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#108

Post by bitwize » Mon, Jan 16, 2023, 11:05 PM UTC

ghaff wrote: post #81

philwelch wrote: post #58

Earlier quoted context omitted.

And losing their GPS to boot

Airplane mode AFAIK (and seems to be confirmed online and in my own experience) does not disable GPS.

Acquiring location via GPS does not require transmitting over the radio, so that alone may well not be affected by airplane mode.

However, most map apps download street map info through the network and cache it for unspecified amounts of time, so unless you use OsmAnd or something, you may lose map data and turn-by-turn navigation.

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#109

Post by ars » Mon, Jan 16, 2023, 11:06 PM UTC

incanus77 wrote: post #105

ars wrote: post #99

Earlier quoted context omitted.

You seem to have a strange blind spot for Apple. Amazon is doing the exact same thing: Only sending to the cloud what it believes to be a wake word. The stuff you are writing is just not true. > This process not only reduces the probability that "Hey Siri" spoken by another person ... Contrast with Amazon Siri is personal on a single phone, while Alexa is meant to reply to anyone in the room. You need to figure out w…

It's true, I have a long Apple history and I do think that Apple means what it says about prioritizing privacy over convenience (and in many cases, accuracy — Siri is quite frustrating at times). They have taken a not-cloud-first approach which causes the quality to suffer for the tradeoff of gaining plausible deniability with storage of personal data and emphasis on user privacy. But I still think Apple's approach i…

> And the fact that wake words will never be validated in the cloud is arguably (I'm trying to make the point) better

This is what I mean by "blind spot", Amazon is not validating wake words in the cloud - it's filtering out REAL wake words that it thinks might be TV words.

Siri on the other hand does no such filtering, so is arguably worse. You think that's it's better, but that's because you have not thought this all the way through.

Re: Why Alexa won't wake up when hearing "Alexa" in Amazon's Super Bowl ad (2019)

#110

Post by incanus77 » Mon, Jan 16, 2023, 11:17 PM UTC

ars wrote: post #109

incanus77 wrote: post #105

Earlier quoted context omitted.

It's true, I have a long Apple history and I do think that Apple means what it says about prioritizing privacy over convenience (and in many cases, accuracy — Siri is quite frustrating at times). They have taken a not-cloud-first approach which causes the quality to suffer for the tradeoff of gaining plausible deniability with storage of personal data and emphasis on user privacy. But I still think Apple's approach i…

> And the fact that wake words will never be validated in the cloud is arguably (I'm trying to make the point) better This is what I mean by "blind spot", Amazon is not validating wake words in the cloud - it's filtering out REAL wake words that it thinks might be TV words. Siri on the other hand does no such filtering, so is arguably worse . You think that's it's better, but that's because you have not thought this…

I am fully onboard with the fact that Siri is the worst personal digital assistant out there.

Will think on the rest, and do some more reading. I appreciate the responses.