This fairly rock and roll sounding header reflects the intense sessions i have been spending in the labs at uni. I have pretty much been the last to leave in the evening along with my colleague Ben Reynhart, getting kicked out by the security guards. I have been plowing ahead with my patcher making great progress. Over the past couple of days i have integrated some key elements into my patcher. The first of these is the implementation of the playback once the user has finished talking. I have used a period of hangover time, the inspiration for which came from reading a paper on ‘Conversational Model Based VoIP Traffic Generation’.
Basically, during speech the voice lulls and dips between words. To avoid my patcher thinking these were the end of a sentence i have created a period of time once the voice drops below a threshold to see if it remains at this level. Once the hangover time has completed and the voice has not risen above this level, the sound in the buffer is recorded and a sample is triggered.
The samples triggered are based on a random selection. A random number is generated between 0 and the amount of files in the directory.
This number is then checked against a list to see if it has been generated already this session to avoid repetition of samples. If it matches a number in the list, a new random number is generated and the same process is applied. If it isn’t in the list, that number sample is then played also adding that number to the list of generated numbers. phew!
I have also added a whitenoise function, which generates distortion on a completely random basis. This is to provoke a feeling of unease in the user, and explore what happens when there is interference in a phone conversation.
Tags: hangover time, idat305, sound, speech samples, white noise