LAB 2: Speech rate

1. The purpose of this lab is to explore the factors that affect speech rate in conversational speech using the Fisher transcripts corpus (LDC2004T19).

2. There is a copy of the Fisher corpus on Harris: /corpora/Fisher_English_Training_Transcripts_Part1/. Read the documentations in the directory docs, starting from the file fe_03_readme.txt. Which file contains speaker information? The transcripts you will be using are in ./package/fe_03_p1_tran/data/trans/. (note: if you are new to unix/linux, here is a nice tutorial for beginners: http://www.ee.surrey.ac.uk/Teaching/Unix/).

3. Write a python program(s) to do the following:

• Calculate the average speech rate, i.e., the number of words per minute, of each speaker in the corpus. Some tokens appeared in the transcripts are not words, for example, ((, g-, [cough], etc. It’s up to you how to treat those tokens.

• Calculate the mean speech rate and its standard deviation against the following factors: sex (male and female), age (it’s up to you how to group ages), and where the speaker was raised (it’s up to you how to group the places).

4. Write a lab report that includes the following:

• Your results and findings on which factors affect (or do not affect) speech rate, and a brief discussion on your findings.
• The technical issues you met and how you solved the problems, for example, the issues of non-word tokens, speaker information errors, etc.
• A print-out of your code and a sample copy of the output of the code.