Audio Detection Program

Time to call upon the vast pool of knowledge of the best tech community on the internet!

I'm looking for a program or app that will allow me to place a microphone that will detect key phrases and spit out prerecorded responses.

I'm looking to stage a very simple conversation with a "ghost" through a radio. The obvious route is to record everything on a time delay but because the event in question has a variety of variables I'm not entirely comfortable that I can maintain a timeline.

Hopefully someone has some experience here and can point me in the right direction. I'm certainly willing to entertain other options though. A wireless remote that will allow me to page through recordings in succession would also work adequately, but it would mean I have to "ask" questions in a precise order. What do we think? Does such an audio detection program exist?

Thanks guys! I'm sure you'll knock this one out of the park for me!

Such high praise of us! It's a shame we have no responses ha.

Such a program sounds interesting. I don't know of any specifically, unfortunately. If you end up with the pre-recorded manual route, perhaps you can have another person on another end operate the play back responses, that way you don't have to go in a set order and the other person can choose responses.

What are you doing this for?

For Halloween this year instead of a boring old party I am hosting a murder mystery of my own design (The prefabricated ones are dreadful). The theme is 1930s Louisiana Seance. It all centers around an old 1930s tombstone radio that has been re-purposed and repaired that will be used to "communicate" with the deceased in order to obtain clues. Linear is certainly a viable fallback plan but wouldn't it be something if I could have the guests ask questions and let them lead instead of having to lead myself because I am the only one who knows the answers in order? As much as I would like to have a man on the inside to control things for me I'd feel terrible asking one of my friends to exclude themselves from the festivities.

It would also be nice to reduce the amount of buttons, switches and pulleys I have to conceal and use.

Oh, that sounds fun.

Text-to-speech apps as a basis comes to mind in terms of the technology for recognizing speech and issuing commands, but not sure where to go from there. Maybe taking that and customizing the responses to whatever you wanted. Not sure if any apps do that now.

As for asking a friend, I guess that would be excluding them but I dunno, personally I find it very fun to be on the acting end. Usually my friend and I, we always do the activities/pranks/jokes together on our friends.

Whatever you do, good luck though!

I can't think of any single programs that will do what you want. But if you combined a voice recognition system (like Dragon) you could potentially set up they key phrases to trigger the recordings to play. But there are many obvious flaws to that, the first coming to mind that if your phrases are aucistically similar you will run the risk of playing the wrong recording at the wrong time. And it may be complicated setting that up, I'm not sure if that software is capable of actually recognizing and comparing when a specific phrase is said to a preset counterpart. I am fairly confident that some of them are designed as a vocal I\O interface, though. I will look into it more after work, but I imagine you would be able to coordinate vocal commands in the form of your phrases and attach them to activate your recordings.

A more rudimentary approach would be to use a CD player or a PowerPoint with attached audio, and just have a remote to transition between then. But that has the issue that, unless its sequential, it will be difficult to make sure you hit the right slide, as you already noted.

You could try to write your own using an open Speech-Recognition-Library.
I know that java has one, maybe you want to give it a try. But I bet there are ones for other languages as well.
If you got your program ready to recognize speech then playing an audio should not be a huge deal.

There are a few for CPP as well as some full vocal models, but I can't remember the names right now. And I assume you want something less complicated anyway, the ones I looked into in the past were not easy to integrate.

As I said, I think a voice based input interface like dragon would be able to do it, but I just got my ass kicked at work so I'm not looking up how to use them right now.

Testing with voice recognition software leads me to believe that this would not be sensitive enough to function as intended. The software frequently tries to interpret white noise or unrelated conversation and makes the whole thing far too unstable for the environment I wish to place it in.

The only other way I can think of is a second person, lest you want to go by a more intricate system of switches. But then mobility is more limited since you need a physical remote or something to control it. I'll ponder some more at work, as it appears I will have time.