It occurs to me that here we all are working away at voice interactive projects but sometimes unaware of what others are doing. I thought it would be interesting to share some examples of projects.
To start the ball rolling, here is an .ogg recording of a sample session with my current dialog manager (Athena, named after the DM in one of Arthur C. Clarke's books). It is a 3.8 MB OGG download and the quality is not very good but should be decipherable. I'm having issues recording a session from both bluetooth input and output simultaneously, so this was created from two independent devices and then merged into one in Audacity. The in and out tracks are not quite perfectly aligned, but it should give an example.
The session is unedited apart from alignment and there is at least one out of context detection.
I'm sure your project is a quantum leap up from this. Care to post an example?
--- (Edited on 2/22/2012 10:47 am [GMT-0600] by colbec) ---
Here is a second example.
In this experiment the dialog manager challenges the user with randomly selected prompts from the grammar. User then enunciates the prompt and the DM evaluates whether it heard what was expected. If true it just says "OK" and moves on to the next. When there is an error, things get a bit more complicated.
Using this method, some of the errors are generated by my failure to understand what the festival voice enunciates. There are ways around this.
Mono channel, about 3 MB, 5 minutes playing time. In this sample there are some deliberate pops generated by the DM to indicate that it is ready for a new sample recording.
--- (Edited on 3/6/2012 10:42 am [GMT-0600] by colbec) ---