So. I understand that if takes a lot to make a genuine "speech recognition" going, and I have read the nice post: https://synthiam.com/Community/Questions/8420 regarding this. My question is not so much "how do I add speech recognition", but rather "how would I make my robot respond to simple audio commands through it own microphone".
First off, I guess Im wondering if anyone has made their robot able to respond to a general "Yes" / "No" audio command (I guess if it could be taught the difference between "E" and "O" that would be sufficient), and if it would work for "most" voices (not only the programmers voice). Anyone done this or know of a good source for info on this subject?
Secondly, there is the Microphone input / hardware issue. If Im not needing my robot to understand what Im saying, but just respond to "E" or "O" (or single clapp / double clapp aso), what hardware would you recommend I try to figure out? Is there such a thing as a WiFi / BT microphone that anyone has made EZ-B connect to? Or would it be easier to upgrade EZ-B with an expansion / shield of some sort? Is there any other sensor out there that is cheap and could either be set to monitor frequenzy (audible) to at least make my robot stop if any high pitched / loud sounding noise is registered?
I want to program my robot on my desktop, but would love for it to have half an ear so that it may be controlled by simle audio inputs once out of the desk enviroment, or maybe I could make it stop whatever its doing by screaming NOOOOO to it
Thank you all in advance!
Claus
The app is caled WO Mic, its an app that you have to install on the phone and on the PC. You can chose to have it work as BT or wifi. On the PC side it is like a regular microfone. I still use EZ- builder on the pc.
Thank you proteusy
A bit of clarification from one of your questions. Although there is a mobile version of ARC that runs on Android, it doesn't support speech recognition, so you will need a PC somewhere, either on the bot or off.
The WO microphone allows you to use an android phone as your microphone for the remote PC. If you mount a PC on the bot, you could just use a wired microphone, but there are a lot of reasons not to do that. Robots can be noisy and can prevent good recognition.
Alan
Thank you Alan. I dont need speech recognition per say. Just an audio-responsive robot.
Lets say my robot will be put out in the woods and it will navigate on its own. It will encounter obstacles and react as it is programmed to. What I want is for the audio to register (periodically or live / streaming) so that if my robot registers a given sound (could be "sounds louder than usual", or "Any sound of some Hz/ pitch" or "Any sound equalling beforementioned that also lasts for more than X seconds" ).
In real life, my robot would be "listening" for (at intervalls or all the time) Bird songs. If it hears something that could be a Bird song, It will stop dead in its tracks and start recording / streaming / logging GPS data. After preset value of some sort (time) it will "reboot" / restart its journey through the woods.
If some scary creature (human) should come along and interfere, It would be nice if my robot could detect the presence of vocal / audio from it (maybe stop, record / stream, GPS log again), but more importantly relay a pre-recorded message saying something like "Hello humanoid. I am XXX and I am a robot sent from a galaxy far, far away to record Bird Songs ... (bla, bla) My creators have been notified of your precense, and a warparty is on its way right this minute (to GPS coordinate XXXXXXXXXX). Please let me continue my precious mission, or I will self-destruct in 10 - - 9 - - 8 - - ..." aso...
Not speech recognition at all. Just audio in to trigger actions.
I saw another post where someone wanted to rig ADC-ports to do something similar, and Ill look into doing that alongside using the most compact PC-based solution I can find.
Thank you for clarifying. My questions are somewhat influenced by my complete lack of understanding of why not everybody wants audio input on the EZ-B. Interference or not. I can think of hundreds of uses of the EZ-B4 that will never see a servo...
I will probably build a small PC and put inside my robot, and I hereby vote for a poll where users could vote for or against audio in jack on the EZ-B5, and a vote for or against some arduino-ish means of loading code onto EZ-B to let it operate without a PC running ARC
The ezb (whatever version) will never operate without a PC or mobile device... It is a a virtual "brick" without connecting to a PC or mobile device... The reason it is so powerful is because of the software that runs on windows or your mobile device... If you don't want a PC on board your robot you will need to be within range of wifi of your robot using a remote PC or mobile device... Plain and simple...
FYI the PC you want to build will have to run a minimum of Windows 7...
If you want something with as little capability and processing power as an Arduino running your code, no reason to have an EZ-B in between it and your servos and sensors, just get an Arduino and learn to write code for it. If you want the capabilities of ARC, you need a computer, not a microprocessor, to run it. That computer can be an iPhone or Android device (within limits), but it is not going to be a single chip solution for many years.
PCs are getting very small. No reason on anything except a micro-bot not to just put a windows computer on board if that is what you want. There will shortly be an expansion board available for the EZ-B that will allow you to connect with USB instead of WiFi, so metal bodies and such become non issues.
Alan
Thank you Richard. I dont mind building a PC into the robot to make it semi-automnous. It will surely make my robot very powerful. Its just a bummer that this makes my robot exactly as stable as Windows is, and will require more power... Wouldnt exactly send a ROVER to Mars relying on Windows stability
An onboard compiled application would be very nice. "Save to EZ-B"-option directly from ARC would make Arduino-users come running also. Maybe some time in the future. Maybe that USB-UART-thing (saw someone discussing it on the forum) will open some new doors?
I guess Windows 7 is ok. Should be pretty stable, and I can add a paralell IQ and sensor array of sorts to send GSM GPS data and record / stream audio directly from the PC over GSM /4G.
I cant find Minimum System Requirement for ARC. Anyone tried running it on Celeron N3050 or 2957U or Atom Z3735F? Anyone measured actual power consumption on a basic intel NUC? I found the MSI Cubi only draws 10W idling and 24W on "full speed". MSI Cubi is i3. If I could run ARC off some lighter processor that would be great.
And thank you all again for your replies to all my rookie questions!
Robots are a fascinating subject, but before pulling the cannons out....
I advise you to learn the EZ-*, perform the tutorials, to really understand what you can do with EZB, what you can't do without more skills.
For example navigation, obstacle detection, sound (not speech) recognition are advanced topics which you will need other software skills (programming).
And even being a skilled programmer, some issues are not easy to implement.
Sometimes new ideas arrive while learning the tools, so i would start first with the existent tutorials.