Is there any program which enables you to talk via microphone and program a chatbot to react with voice answer

My Twitter bot @chatbotix just picked this question up from answers.yahoo.com, and I thought the subtleties worth re-posting here.  I don’t think “program” is the right word, “teach” seems more reasonable.  So, are there any bundled platforms, or mobile apps, that can both be taught by voice and respond by voice?

A while back, 8pla was working on a chatbot with a voice interface, but it only worked in Google Chrome. I’m sure that there are others out there, but I’m not currently aware of them.

Now, you have to “roll your own”. You need; speech to text, a chatbot engine, and text to speech.

In the old days, you could do it on a Windows system with Microsoft Agent. One of my earlier chatbots (Merlin) used to do this. You could still if you are a programmer using c#.

Web based speech input is included on Apple and Chrome platforms.

Hopefully, as standards arrive, it will get easier in the future.
http://www.w3.org/TR/speech-synthesis/

TTS
http://dyslexiahelp.umich.edu/latest/text-speech-and-other-plugins-web-browsers
https://developer.chrome.com/apps/tts

Zabaware has some of these capabilities.
http://www.zabaware.com/reader/

as does Guile3d
https://guile3d.com/en/

i am looking at the posabiliy of doing the text to speech bit, using chat script for patern recognition.

don’t get excited, im still looking at if it is going to be doable

HTML5/JavaScript supports both text to speech, and voice recognition, so that is probably the best option for web.

For mobile the Android API, and I assume iOS support tts and voice rec.

For JavaScript see,
http://updates.html5rocks.com/2014/01/Web-apps-that-talk—-Introduction-to-the-Speech-Synthesis-API

and,

http://shapeshed.com/html5-speech-recognition-api/

Chrome used to have a x-webkit option for voice rec in an input, but now uses the HTML5 javascript standard.

Only Chrome supports both of these so far that I know of.

BOT libre supports both tts and voice rec in Android and web.  We use the Chrome HTML5 API for voice rec (only works on Chrome (click listen checkbox to activate in a bot chat session)), and we generate our own tts on our server using the Mary Java tts project.  We also publish our own free tts JavaScript API.

http://botlibre.blogspot.ca/2014/11/add-speech-to-your-website-using-bot.html

Text to speech is very well established and I consider it a solved problem.

Voice rec, not so much.
I’m not sure if the problem is hardware or software.  Replicating the ear may be the hardest part of voice rec.  I’ve tried, Chrome, Android, Google, iOS voice rec, and none was that good I far as I am concerned.  Perhaps I just don’t annunciate properly, or tend to mumble.  I think we still have a ways to go before voice is solved.  If your going to try voice rec, having a very high end microphone is a must.

Voice recognition won’t be solved for a while. You can only get so far with wave pattern matching and statistical probabilities of words succeeding other words. To improve beyond that we need A.I. that takes both common knowledge and context into account to decide which of several possible translations adds up to making sense. And that is about the same challenge as building a human-level A.I. in the first place.
To improve your immediate experiences with current voice technology though: Talk like a robot.

Google’s voice rec does take context into account, at least the Google search voice rec does.  If you go to Google search from Chrome and click on the microphone and start talking it changes what it thinks you said after each word, it is pretty cool.  It can get your first word wrong, but then after you second or third it figures out what you meant to say.  So they are using context and statistics on which words go with which other words, and what people (like you, as they keep stats on you as well) normally search for, maybe were closer than you think.

The voice rec API in Chrome and Android doesn’t seem as advanced though, seems to be specific to Google search.

Not sure if this is still an active thread, but you can integrate speech recognition very easily with the Microsoft Bot Framework and the Bing Speech API:

https://dev.botframework.com
https://azure.microsoft.com/en-us/services/cognitive-services/speech/

Hi Jasson,

Even before Microsoft Bot Framework (which I have not tried yet), Microsoft was always excellent for voice recognition built into Windows, in my opinion. Thanks for recommending the Microsoft Bot Framework and the Bing Speech API. Seems they just discontinued support on a bunch of technologies at Microsoft.  Personally, I am going to miss VBScript. It was never as widely adopted as JavaScript, but VBScript was a powerful beast.

∞Pla•Net - Sep 23, 2017:

...but VBScript was a powerful beast.

Powerful, yes, but also a HUGE potential security threat, since it could be used to gain access to both the filesystem, and to critical low-level OS functions. For a time during the early to mid ‘naughts’ VBScript “virii” (they weren’t really virii, but they had the same malicious intent, so…) were taking up the majority of the “top ten” worst threat slots on several AV sites. things have changed since then, of course, but the potential for VBScript-based malware is still a possibility. Like you, I like VBScript, but I think the risk is greater than the benefit. Just sayin’.

∞Pla•Net - Sep 23, 2017:

Seems they just discontinued support on a bunch of technologies at Microsoft.

Yeah, Microsoft is famous for that, LOL.  They just released the Bot Framework last year, and they already have a lot of traction with it. Plus, they are getting ready to release a lot of cool enhancements to the Bot Framework (like Conversation Designer), so this one’s probably not going anywhere anytime soon.