Amazon Is Discontinuing the "Do Not Send Voice Recordings" Feature on Echo | Hacker News Reader

Amazon Is Discontinuing the "Do Not Send Voice Recordings" Feature on Echo

resetera.com

182 pointsnickthegreek28 comments

I deleted my Amazon account some time ago. It seems that with subsequent updates, they must have reactivated recording by default again. I discovered 2.8 GB of voice recordings, and creepily, not all of them were initiated by the designated trigger word. However, with the provided command detected JSON, you can analyze the recordings to identify words you may be mispronouncing.

truly curious what sort of cost benefit analysis results in: I am comfortable with inviting a rapacious corporation into my private life.

Im getting old and find all features provided by voice assistants very much unnecessary, but perhaps younger generations cannot live without it?

Those features did not exist say a decade ago, so cannot live without is likely to be a dramatisation of a minor inconvenience.

It's sort of a shame the industry is in the state it's in, because I personally know a couple of folks who have disabilities that make something like an Echo truly a godsend.

There are a lot of things that get much, much easier from an accessibility standpoint if someone can operate a device by voice, instead of having to maneuver next to a switch, panel, or button.

Getting next to the button can be much more difficult if it means getting into a wheelchair, fitting your prosthetic, or having to find help because it's positioned too high or in an area that's too narrow.

Additionally, you can say a single phrase and turn off all lights in your dwelling, regardless of room. Minor win for able-bodied folks who might take only a minute or two to walk the house, big win for someone less capable.

---

I'm technically capable enough to have removed Amazon/Google anyways with HomeAssistant in combination with WIS (https://github.com/toverainc/willow-inference-server) and Willow autocorrect (https://github.com/toverainc/willow-autocorrect).

Which gets me most of the things I cared about from an Echo without ever having anything leave my LAN. But it's a real PITA the setup.

There is basically zero middleground in this industry between "Amazon/Google have a permanent microphone that listens to everything you do" and "You have to manually flash ESP32s and configure servers".

Makes sense. As often happens, it is the most vulnerable ones who are being taken advantage of.

Which mic/speaker do you use with WIS?

https://www.adafruit.com/product/5290 (Original - I don't believe it's made anymore)

and

https://www.adafruit.com/product/5835 (Updated version of the above)

---

Neither are particularly awesome as far as devices go, but they're cheap enough for what they are, and they're the targeted hardware for willow (https://heywillow.io/hardware/)

You won't be playing music with them, and the audio pickup isn't as good as an Echo. But they work and my family/folks who enter my home don't get recorded automatically by Google/Amazon.

It's very, very similar hardware to what Home Assistant is trying to use (ESP32 in a box with some mics).

I had a parallel thought when a loved one was in the hospital. I noticed the nurses were so busy that small requests like an extra blanket were often forgotten entirely. How incredible would it be to have a voice system that would build a task list and capture other notable bits of information that the hospital staff could review on an integrated device (like a phone)?

Sadly, I don't see this ever happening due to the insane number of regulations you'd have to satisfy and the major players in this space only doing it to collect PII.

Good use case for Siri AI with iPad in kiosk mode, big button for each task for visibility at distance, pressed by nurse when done, notification sent to loved ones.

I find hospitals to be somewhat clueless when it comes to simple things like what you’re describing.

When a relative of mine got hospitalized for a laryngectomy, post op she obviously couldn’t talk since her voice box was gone.

She was either expected to learn to speak with her oesophagus ( doable but many people fail and it takes months ) or … remain mute.

That’s when I introduced her to the wonders of text to speech. Soon after she was the star of the department, which had never considered this !

> zero middleground in this industry.. manually flash ESP32s and configure servers

Market opportunity for local "home electronics contractor" network.

"I have nothing to hide, so I have nothing to fear." Or "I don't know any better"

How is the open-source voice assistant space these days? Is there a self-hosted alternative to Alexa these days?

Home Assistant is working on one

https://www.home-assistant.io/voice-pe/

Yes. I use https://heywillow.io/ and nothing ever leaves my LAN.

It's entirely open, and relatively hackable - but also a pain to configure and requires 1 machine with a GPU (several years old is fine, I'm running mine on a nearly decade old Nvidia GTX 1080) and a couple satellite ESP32 devices (Espressif ESP32-S3-BOX-3).

Home Assistant is also doing a more integrated version, but last I looked it was still "cloud-only" which defeats the purpose for me, since I was interested from a privacy standpoint. It does look like they finally released a local preview (at least for some languages) but I haven't yet tried them. By default, it still looks like they're using Google for TTS, which is a hard "NO!" from me.

I've been pretty disappointed that they really seem to want to drive monetization with their voice features, and force people into the HA cloud services, but again - I haven't tried the latest release so take my frustrating with a grain of salt. They made a lot of noise in early 2023, and late last year still only supported cloud options... I was annoyed.

Unsuprising. Of course Amazon will do this. They need to.

Just another reminder that *any* company marketing about "privacy" in their product keynotes will always do the exact opposite.

Best example of this was Mozilla. Preaching "privacyin Firefox browser and now is deploying ad-tracking into their browser (after acquiring Anonym that sells your data to advertisers "privately").

do you think it's feasible to create a Gen AI version of Alexa completely locally? unsure what exactly you're trying to say.

HomeAssistant already does

Hmm yeah but it needs 30 seconds to give an answer. And that's with the LLM running on a GPU with HBM2 memory.

yes. that was the original design actually. and it worked. latency was a bit much though.

It's not "Gen"erative, but is still a Transformer, and is kinda OK: https://github.com/openai/whisper

(I'm not too impressed with the error rate, but OTOH Alexa and Siri mishear me fairly often as well, so Whisper may be "good enough").

its feasible. if you want to try yourself you can start with the april-asr package.

There are tools to make your own.

Huh I didn't know that existed. So I could have used Alexa all this time.

But now it's going away anyway...

And we definitely know it did exactly what it says...

More discussion: https://news.ycombinator.com/item?id=43367536

So on one hand, sure, Amazon isn't trustworthy.

On the other hand, the big question with Echo/Alexa has always been "is it listening to me all the time?". My understanding is that people have snooped the devices and they appeared to only send data when the wake word was used. Now... does that mean they are not recording all the time and only sending data during the wake word window? I do not know.

However the above point is different. I honestly didn't think that they were generating the text on device to this point, so this is news that they ever WEREN'T sending voice to the cloud. However, I think a lot of people are going to get these two things confused and basically nope out of the Alexa ecosystem.

Reply on news.ycombinator.com