Update on DIY Amazon Echo

An update to this post
I have two ringer phones in always listening states now. One in my kitchen, the other in my living room. The living room phone is hard wired to speakers and uses the phone's mic/receiver to listen. The kitchen phone is largely novelty at this point due to Bluetooth issues which I will explain at the end.

Since the last post, I installed the apps Utter and Commandr and have been messing around with them to create a more fluid Echo-like experience. Both allow more device-specific features than what Google Now offers. One can control volume and open apps on the phone using voice commands.

Utter allows specific voice-wake-words. It essentially works within its own world, but can interact with Google Now and Google servers when it can't figure out what you're asking (or if it needs the internet, like asking about the weather). I set my wake word as Alexa, and with the cheap ZTE phones "she" can typically hear me*. If the TV is on too loud, she may not hear, or get confused from ambient noise. Utter has a pre-set list of commands that it recognizes, but you can also create your own or link Taskr to Utter. It also is always listening if the screen is off.

Commandr integrates into Google Now. So for this app, one still would use "Ok Google" to trigger Google Now. Commandr listens within within Google Now, so if I say "Volume Down", Commandr kicks in and would lower the volume. If the voice command isn't in Commandr's knowledge-range, it doesn't intervene and Google Now continues.

I found that OpenMic+ will goof up Utter, so OpenMic needs to be uninstalled or disabled to get Utter and/or Commandr to work.

Right now, Utter and Commandr together are the closest to a DIY Echo one can make. Although Utter alone is ideal. Tasker is next on my list.

Updated May 4, see below. Using a Bluetooth speaker that has a mic results in some screwiness. Music will come through the speaker, but yelling the voice commands seem to confuse the phone so that it will use the mic in the phone, stop the audio, and then play havoc as to where the TTS results will come from. Sometimes the voice results come out of the handset, sometimes out of the speaker. All in all, this is the next area of focus, because if I can have a lone speaker with a mic in it, and do most assistant-level stuff through it, then we're in business
UPDATE MAY 4
Uninstalling or disabling Commandr allows for a Bluetooth speaker/mic combo to act as an always listening device. So now, only Utter is running in always listening mode. The Anker speaker/mic is paired to any one phone, but the mic is now always listening, with all audio through the speaker. When Commandr was running, it was also always listening in addition to Utter, and confused things to where I never knew where audio was going to come out. Utter wanted to use the Anker, but Commandr kept trying to deliver Google Now commands through the phone speaker.

*Utter allows for a variety of text to speech engines and personalities. I use the default female US Google voice, but you can download voices from IVONA on the Google Play store and use them. I used the female British one, but it was too slow to respond