14

See these instructions. Create an AWS developer account & AWS account. In the AWS console Create a lambda function. Include in the lambda function some code that will access the API. This can either be python or java or node.js. Here is a python script. Change modify_state to be either 1 or 0 import urllib2 def modify_state( port, state, token): ...


11

The Google Home and Amazon Echo use microphone arrays to enhance 'far-field recognition' (i.e. recognising your voice from a reasonable distance with good accuracy). The Echo uses a 7-microphone array (image from iFixit, with the microphones in green) and the Google Home uses a 2-mic array (iFixit; in yellow). Amazon's 7-Mic Array is open for developers ...


9

You could try Raspberry Pi with .NET Core. Microsoft has published decent automation libraries, including speech recognition. This is of course more hands-on, but combine it with other sensors and software, and you could do some pretty cool stuff. Microsoft also utilize their speech recognition libraries in their Bing APIs, so you could potentially use the ...


8

In the US, you can use the Voice Training feature in the Alexa app. The option can be found by going to the Alexa app, selecting Settings, then Voice Training, then Start Session. You will then be instructed to speak several phrases to help train your device to recognise your voice. Amazon provide the following instructions: During a Voice Training session, ...


8

If you have an Android phone, try to use the speaking option of the keyboard or try "Ok Google" and ask something. I suppose Google uses the same speech recognition engine on all its platforms...


8

Aurora0001's great answer got me doing some more research and I found some really good information on a lot of mic arrays, including benchmarks. Seeed ReSpeaker Mic Array Conexant 4-Mic Development Kit Microsemi AcuEdge MATRIX Creator MiniDSP UMA-8 PlayStation Eye (Tonor Stereo Condenser Microphone) medium.com did some awesome benchmarking on these. For ...


6

I've heard of two open solutions working on Raspberry Pi : Jarvis (no kidding) S.A.R.A.H. I didn't test them but I suppose you only need an Internet connection to initialize them.


5

Not an Alexa specific answer, but look into support for soundex and similar phonetic hashing systems in your platform and/or database. For example the MySQL database has a soundex() function that can be used for this. BMPM is another algorithm supported out of the box by Apache Solr/Lucene, along with a number of others. https://lucene.apache.org/solr/guide/...


4

Google home can be trained by saying several times the key word Hello, Google. I am not sure how this takes effect in different pronunciations, but for now this is the only way to teach GH anything. Train it again: unfortunately you have to unlink your ID that's having trouble and then add it back in (scarier than it sounds). It's over by the devices at the ...


4

Basically what you can do for training by yourself is to retrain the device for your voice. You'll say consecutive times the wake words as they are prompted by the app. I found a post stating you should tell the wake words all around your home to get different conditions better taken in care. My claim is that the same would apply if you imitate little bit ...


3

Google has now announced how to set up multiple accounts on the Google Home: Click the 'multi-user is available' card in the Google Home app A list of devices will pop up. Find the correct one, then click "Link your Account" A wizard will then guide you through the steps of adding a new user. To train the device to recognise your voice, you have to say "OK ...


3

This interesting and engaging Youtube video by WIRED shows 8 People Test Their Accents on Siri, Echo and Google Home. The test is based on 4 questions, asked by each person, to each device under test. Google Home emerges better among the 3 devices at understanding variations in accents followed by Siri and Echo.


1

What you are looking for is called oAuth. It's a standard for issuing authentication tokens authorising services to act on behalf of users in a controlled way.


Only top voted, non community-wiki answers of a minimum length are eligible