Google Home and Amazon Echo

[Guest Alphonsox]

Both Google and Amazon have recently released clients for their respective voice control platforms (Echo, Home) on the Raspberry Pi. This provides a fix for the big misgiving I have always had with these devices - The Always-On (or software controlled) microphone. Effectively you can now build a voice control device with a physically switched microphone connection and be certain no-one is listening in when they shouldn't be - I'm thinking a big red push-to-make mushroom button.

 

I've spent a few hours playing around with the browser based Echo interface (echosim.io) and see enough possibilities to investigate further. It's clear that customising the experience by writing personalised Apps/Skills/Actions is going to make things a whole lot more useful, especially if integration with IFTTT (or similar) can be achieved.

 

Does anyone have any experience of these devices (good or bad) and has anyone played with coding for these things ?

 

 

[Jeremy Harris]

Being able to close these systems off is good news indeed, as it seems clear that the always-on mic does collect and record data not directly related to a command, hence the recent trial where Amazon were asked to release audio recording data as evidence as to what had happened in a room immediately before a crime was committed.

 

How are Google and Amazon going to make money from a closed loop system though? 

 

Their business model is primarily based on data sales, in one form or another, and a closed system isn't going to pass them any data to profit from.

 

I had a look a couple of months ago at some DIY options, like Alexa Pi, but all these really did was allow a RPi to act as the front end client and pass data back to Amazon servers for processing, AFAICS, so there's the same issues with data privacy as there are with the Amazon product.

Edited July 2, 2017 by JSHarris

[Guest Alphonsox]

These are in no way closed loop systems. From what I can see  the data being sent to Google or Amazon/Bing is exactly the same as if I had turned on the Laptop and typed the search in on a search engine. For everyday information I don't have a problem with the search providers recording access and hence providing a more targeted response. (If I want privacy I use a VPN and/or DuckDuckGo). What I do have an issue with is information being passed and recorded without my knowledge which is what these devices seems to do by default. The ability to use a physically switched microphone lets me control what information is sent which makes a big difference to my willingness to give them a try.

[Jeremy Harris]

I get you, I misunderstood what had been released.  If it's just a specific instruction to a search engine, with the mic under your direct control, then that sounds a reasonable compromise.  Like you, I have an issue with the data collection that these systems do without you being aware.  The idea of having an always-on mic in the house seems just a bit too much like Big Brother (in the 1984 sense) for me. 

 

With a mic button, the system would presumably work a bit like the way the speech interface in my car works.  That has a "push to talk" button on the steering wheel, which activates the speech recognition system.  Frankly the Toyota system isn't great, and it's usually quicker to just reach across and tap commands directly on the screen, but from the adverts, the Google and Amazon systems seem to work a lot better, presumably because they can quickly access a massive amount of data on their respective servers.