716: Final project, assignment 2 due by Friday 12/06/19, 11:59PM
What you have done already (HW due 11/22/19)
In your Dropbox directories in our shared directory:
- Determined f0 ranges for individual speakers for your language and written to a spreadsheet file
- Run VoiceSauce analyses in Matlab on all of your language files
- Checked over VoiceSauce measurements for reasonableness, e.g., formant values ranges, f0 ranges, and made notes about this
- Written the measurements extracted from VoiceSauce using
output to text in Matlab to a textfile, with 9 sub-segments per text grid interval
- Read the VoiceSauce output measurements text file into R successfully and produced some plots of different measurements in R notebook
Next up: creak space parameters (class time review today)
- Review: UChicago presentation
- Recall Garellek etc. proposal of relevant parameters for creak space
- Constricted voicing/spectral tilt: H1-H2 (also, H1)
- Noise/irregular pitch: harmonics to noise ratios (and CPP)
- Low pitch: f0
- Two issues:
- What parameters discriminate between the "creaky tone" category and the "non-creaky tone" category?
- What parameters crop up within the "creaky tone" category as delineating "types" of creak?
Next up: FDA analysis (due by 11:59PM, 12/06/19)
Already has been implemented previously by me, see fda/ directories in your language directories. But now you do it!
- You will need to "port" the Snw file I wrote into a new R notebook
- You'll need to modify the code for your own data files
- You'll need to comment the code to demonstrate your understanding of what's going on in the code
- You'll need to write up your notes on what's going on in terms of our issue about discriminating between the creaky tone and non-creaky tone category in the R notebook.
- Warning: random forests take a long time to run. If this is a problem for your machine, ignore for now.
What's due: completed R notebook, and html file output: I should be able to open your R notebook in RStudio from your Dropbox folder and just run it with no problem. You will present your findings in class on Tuesday 12/10/19.
Coming up
- On Thursday 12/05, we will discuss:
- Methods for analyzing the FDA data, including clustering
- How to address our second issue about sub-types of creak
- GAMMs
- Homework due
- We will also begin to divy up paper writing (remember, paper is due 12/20!)