AUTOMATED CHECKS

check_dictionary

Checks that each word in the transcription files is also present in the dictionary

check_energy

Produces the overall signal to noise ratio of each conversation input, the noise floor of each, and each utterance in that conversation for which the boundary lies in a region where the energy is higher than the noise floor by more than the specified deviation

check_silence

Determines the length of the silence-only utterances in a transcription file and flags those that are too short

check_bounds

Insures that the start time of every utterance or word is equal to the end time of the previous utterance or word. it also makes sure that the end time of the last utterance or word is equal to the size of the file up to six significant digits and that the start time of the first utterance is zero.

utterance_hist

Determines the length of the non-silence-only utterances in all transcription files and builds a distribution reporting on such information as the mean duration, variance, std. dev, number of utterances within 1, 2, 3, 4, and 5 std. devs of the mean, etc.

check_speech_rate

Determines the speech rate of every non-silence utterance and reports statistics for this information. It also flags utterances with abnormally high or low speech rates