The $10 Echo • Hacker News Board

Re: The $10 Echo

#111

Post by notlisted » Mon, Jan 04, 2016, 12:47 PM UTC

goldenkey wrote: post #68

cthalupa wrote: post #63

Earlier quoted context omitted.

>If some update changed this, how and when would you figure out The echo traffic always flows through your internet - it has no 3G/4G hardware, for example. If you want to do a pcap with tcpdump or similar, you can. When I first received mine, I set up a packet capture on my linux firewall/ips/nat device and looked at traffic outbound to the internet from the echo. If I was particularly paranoid, I could leave this r…

"Honestly if you are worried about this sort of X, you shouldn't live in a [subset of first-world country utilities.]" Ad hominem. C'mon, I expect better specious arguments on Hacker News.

Feel free to replace the word 'you' with 'one', so read "If one were truly worried about spy devices, one should really consider dropping their cell phone in the toilet."

Other commenters have provided enough detail on what is currently a fact with cell phones microphones. Ignoring voice, your location (via GPS and/or cell tower), browsing and calling history, etc. etc. is recorded and transmitted to third parties constantly and not just google either, see http://www.bbc.com/news/technology-34732514

Re: The $10 Echo

#112

Post by pikzen » Mon, Jan 04, 2016, 3:56 PM UTC

Houshalter wrote: post #96

jimrandomh wrote: post #33

I'm not comfortable with the idea of everyone having far-field microphones in their homes. Cell phone spying is bad enough. If these become widespread, then ten years from now instead of a political debate about call metadata it'll be a debate about literally every word ever spoken .

I don't see how it's different than a smartphone. Everyone carries with them a little microphone and GPS now. Who cares about using voice recognition in the home? And the real voice recognition should be turned off 99% of the time. It should just passively listen for a keyword before using any cloud service.

>I don't see how it's different than a smartphone. Everyone carries with them a little microphone and GPS now. Who cares about using voice recognition in the home?

"You already have the flu, why worry about getting gastroenteritis?"

>And the real voice recognition should be turned off 99% of the time. It should just passively listen for a keyword before using any cloud service.

On tools like OP's link, or Jasper, etc. you have the source and you can verify that it's actually not sending out data permanently. With the Echo, you're just hoping Amazon didn't do exactly that. You could watch the network activity, but that would just tell you that some data is sent back, sometimes.

Re: The $10 Echo

#113

Post by dharma1 » Mon, Jan 04, 2016, 4:24 PM UTC

dperfect wrote: post #18

One of the more impressive things about the Echo is its far-field mic. I've used a lot of different voice recognition technologies, but Echo blows everything else away in terms of being able to speak in a normal voice from the other side of the room and have it still understand you. The actual intelligence behind the service may be lacking (or at least behind some of the competition), but for a general-purpose home a…

Agreed. It's an array with 7 MEMS mics, which allows the Echo to do beamforming so they can essentially focus on the speaker while rejecting most of the other sound.

There have been advances in deep learning based speech recognition to reject noise - http://usa.baidu.com/deep-speech-lessons-from-deep-learning/ for instance, but clean audio source helps massively, especially if you are trying to talk from any distance instead of being right next to the mic.

I haven't seen open source libraries for state of the art RNN/LSTM based speech recognition, anyone know if they exist?

MEMS mics have been taking over electret mics recently, they are way smaller and have better SNR (at equivalent size).

I'm building a 4 mic array at the moment, tetrahedral/ambisonics mic for VR use, the mics themselves are very cheap. Mouser and Digikey stock them - http://eu.mouser.com/Sensors/Audio-Sensors/MEMS-Microphones/...

Haven't looked into the software/beamforming part yet - does anyone know if there are open source libraries for that?

Some MEMS mics can also do ultrasound range, opening up pretty interesting possibilities for cheapo LIDAR type echolocation with a mic array - but that's for the next project :)

Re: The $10 Echo

#114

Post by kovacs » Mon, Jan 04, 2016, 8:23 PM UTC

praccu wrote: post #76

kovacs wrote: post #47

This is a pretty cool experiment. As someone that recently worked on an app that was doing speech to text in near and far field, calling this a $10 echo is like calling a Honda Civic a $15K Ferrari. I'd love to see the video of the author speaking to this microphone from 10' away, or even 5' away in a room with no padding. The Echo's ability to do voice to text is like nothing that's ever come before it based on my e…

http://www.cobaltspeech.com ?

Yes sir

Re: The $10 Echo

#115

Post by kovacs » Mon, Jan 04, 2016, 8:24 PM UTC

sammachin wrote: post #53

kovacs wrote: post #47

This is a pretty cool experiment. As someone that recently worked on an app that was doing speech to text in near and far field, calling this a $10 echo is like calling a Honda Civic a $15K Ferrari. I'd love to see the video of the author speaking to this microphone from 10' away, or even 5' away in a room with no padding. The Echo's ability to do voice to text is like nothing that's ever come before it based on my e…

I totally agree, hence my comments about artistic license in the title! One of the issues right now is that people kinda assume that Echo and Alexa are the same thing hopefully as more 3rd party devices come to market using Alexa the difference will emerge. I'm thinking of doing a follow on post explaining the differences. Also there's not really a generic name for the type of appliance that the Echo is like smartpho…

From your lips to Amazon's ears, they sure hope so :)

Re: The $10 Echo

#116

Post by lsaferite » Mon, Jan 04, 2016, 9:34 PM UTC

Daneel_ wrote: post #81

freyr wrote: post #41

Earlier quoted context omitted.

> a horrible job providing answers to even simple questions. If you think of it as a device that performs specific actions (i.e., plays audio, gives a weather forecast, sets a timer, etc.) via an voice interface, rather than a general intelligence that can answer random questions, you're going to have a better time.

THIS. It baffles me how people expect devices such as the Echo to understand every nuance of human language, then get annoyed when it can't answer your overly-complex or vague question. Voice interfaces for me are always reduced to clearly enunciated and simple commands such as "set timer 10 minutes" or "play album Anjunadeep 4". You'll have a much better time than saying "alexaplaymesomealbumsbyaboveandbyond"

Well, in fairness, Google does a much better job than Alexa at 'doing what I want'

Re: The $10 Echo

#117

Post by lsaferite » Mon, Jan 04, 2016, 9:44 PM UTC

toomuchtodo wrote: post #108

squeaky-clean wrote: post #35

Earlier quoted context omitted.

Google's question parsing is ridiculously good. You can also do neat things like say "When I get to work, remind me to email John" and you'll get a notification when you arrive. I don't have an Echo, but I imagine it can't do such contextual things. You could get a dedicated Android phone/tablet, and leave it always plugged into a charger somewhere. I just tested it on my Phone (LG G3 Android 5.0.1), and I'm able to…

I love my Nest Thermostat, and I know they've got "Protect" as the smoke/CO detector. I'd LOVE if Google would release an "Assist" product that was the Echo form factor powered by Google's agent engine, and I could mount it on the ceiling in each of my rooms similar to how smoke detectors are (hard wired to mains with battery backup). Too much to hope for I suspect.

That would be both awesome and terrifying.

I have an Echo and the mic array is fantastic. And the idea of having one in every room for localized interaction is great. But the idea of such a good mic in every room is scary if you consider bad and/or state actors.

Re: The $10 Echo

#118

Post by a3_nm » Tue, Jan 05, 2016, 9:46 AM UTC

cthalupa wrote: post #63

a3_nm wrote: post #36

Earlier quoted context omitted.

> The traffic from the Echo has been studied extensively. Detection of the wake word is hardware-based For now, maybe. If some update changed this, how and when would you figure out? I'd say it's not entirely silly to be worried about something in your home which is controlled by a third party and could turn into a spying device at any moment, even though it seems not to have become evil yet.

>If some update changed this, how and when would you figure out The echo traffic always flows through your internet - it has no 3G/4G hardware, for example. If you want to do a pcap with tcpdump or similar, you can. When I first received mine, I set up a packet capture on my linux firewall/ips/nat device and looked at traffic outbound to the internet from the echo. If I was particularly paranoid, I could leave this r…

> traffic outbound to the internet from the echo

A paranoid fear would be that the Echo could be recording additional traffic (based on special stealthy trigger words maybe) and sending that traffic later, in conjunction with regular traffic recorded for the regular trigger word, so as to escape detection. It does sound a bit harder to pull off, I just wanted to point out that spying on the traffic won't get you perfect security if (as I expect) the traffic is encrypted.

> Honestly, if you're worried about this sort of thing, you shouldn't own a cell phone.

Fair point, but as further comments pointed out, taking one kind of risk doesn't mean you're OK with taking more of the same kind of risk. :)

I already take fair precautions against my cell phone, but I have to concede that, with radio firmwares being proprietary, it seems true that any cell phone nowadays is as risky as the echo.

Re: The $10 Echo

#119

Post by dharma1 » Tue, Jan 05, 2016, 9:51 AM UTC

squeaky-clean wrote: post #35

notlisted wrote: post #15

I love my echo (alexa), especially the convenience of "no button but wake word" when I have my hands full in the kitchen. That said, google voice is so much better at parsing my questions that I'd pay a bunch of money for another standalone device with wake word which would integrate with with google voice. Any chance of hacking that, or linking your device with both? (Ask Amazon / Ask Google)

Google's question parsing is ridiculously good. You can also do neat things like say "When I get to work, remind me to email John" and you'll get a notification when you arrive. I don't have an Echo, but I imagine it can't do such contextual things. You could get a dedicated Android phone/tablet, and leave it always plugged into a charger somewhere. I just tested it on my Phone (LG G3 Android 5.0.1), and I'm able to…

I've got an old Android phone always plugged in, connected to a THX system on top of my fridge/freezer, for doing things like automatically playing music in the morning. Exploring what other things I can get it to do with Tasker atm.

OK Google works, but the mic really isn't as good as the array on Echo. I've been looking into external mics for Android - will test with simple omni mics first but ideally an external array mic would be the best choice.

Re: The $10 Echo

#120

Post by jononor » Tue, Jan 05, 2016, 11:03 AM UTC

dharma1 wrote: post #113

dperfect wrote: post #18

One of the more impressive things about the Echo is its far-field mic. I've used a lot of different voice recognition technologies, but Echo blows everything else away in terms of being able to speak in a normal voice from the other side of the room and have it still understand you. The actual intelligence behind the service may be lacking (or at least behind some of the competition), but for a general-purpose home a…

Agreed. It's an array with 7 MEMS mics, which allows the Echo to do beamforming so they can essentially focus on the speaker while rejecting most of the other sound. There have been advances in deep learning based speech recognition to reject noise - http://usa.baidu.com/deep-speech-lessons-from-deep-learning/ for instance, but clean audio source helps massively, especially if you are trying to talk from any distance…

One of the simplest beamforming algorithms are delay-and-sum. This explanation has example code in C: http://www.labbookpages.co.uk/audio/beamforming/delaySum.htm...

PS: interested in your project, anywhere I can follow it?