Output files

After a completed folder analysis the working directory will contain the following folders:

• csvs
• formants
• formants_edited
• formants_winners
• images_comparison
• images_edited
• images_winners
• infos
• infos_aggregated
• processed_data
• regression_infos
• sounds

The folders contain information related to the stage of analysis when they are created. The data created at each analysis step is presented below in this order.

Track formants

In this stage a file will be created called 'file_information.csv'. This file will contain information about the sound files being analyzed. If you extracted sounds form a TextGrid using the functions in Fast Track, then this is made for you already. This file contains 5 columns: file number, file name, label, group, and color. The last three are just for plotting. If you set these up to contain colors or IPA labels, Fast Track will use the information in this file to make nice plots for you when the folder analysis finishes.

formants

The formants folder contains the Praat formant analysis objects saves as plain text files. The naming convention is 'filename_N_.Formant' where 'filename' is the name of the wav file being analyzed. The N in the file name correspond to the analysis step, and the extra '_' after the 'N' is just to make my life easier. This folder will contain (total number of files) x (total analysis steps) files so you hay have a lot, but they will usually be tiny (<50 KB).

infos

The info folder contain the log files detailing the analysis carried out for each file. The folder and log files are created not but are updated as the analysis proceeds. The structure of the log files is explained here

Select Formants

At this stage a file will be created called 'winners.csv'. This file indicates which is the best analysis according to Fast Track. The 'winner' column indicates the best single analysis for all formants, and the columns named 'fN' indicate the winner for formant N. I don't think the edit column does anything, I should probably get rid of it.

images_comparison

The images_comparison folder contains images comparing the analyses for each sound. The naming convention is 'filename_comparison.png'. These files tend to be 1-2 MB each. If you have a large amount of files, (> 2000 ?) you may not want this much data for this analysis. I am going to add a function that generates these after for a subset of analyses. For an explanation of the way the figures should be interpreted, please see this page.

Get Winners

At this stage Fast Track reads from the 'winners.csv' file. You can overwrite it manually and run (or re-run) this step without re-analyzing the sounds.

csvs

The csvs folder contains csv files representing the 'winning' analysis in tabular form. Estimated formant frequencies are always written out. These columns are named 'fN' for formant number 'N'. Time (expressed in seconds) is also always written out. The user can select which other information gets written out by default by following the process [outlined here])(Options). You can choose to output or omit:

• formant bandwidths. These are named 'bN' for formant number N.
• predicted formants. These are named 'fNp' for formant number N.
• pitch ('f0').
• intensity.
• harmonicity.
• normalized time (normalized time = (end time - start time) / duration).

The setting determined with respect to csv output are saved in a text files and remembered across sessions.

formants_winners

The formants_winners folder contains copies of the 'winning' analyses as read from the 'winners.csv' file. The naming convention is 'filename_winner_.Formant'.

images_winners

The images_winners folder contains images of only the winning analyses from the 'winners.csv' file. The naming convention is 'filename_winner_.Formant'.

infos_aggregated

I use the files in here to speed up some other processes. This info may be useful to people but only to extend Fast Track so I will get to it later (but email me if you really want to know).

regression_infos

This folder contains detailed information about the regressions carried out for each file. The naming information for each file is 'filename.txt'. Inside the file, the first 3 lines give basic information about the analysis. The file then gives the following repeating information for each analysis step:

• residual error of regression: F1, F2, F3 (F4)
• F1: coefficient1, coefficient2 , coefficient3 , ... , coefficientN
• F2: coefficient1, coefficient2 , coefficient3 , ... , coefficientN
• F3: coefficient1, coefficient2 , coefficient3 , ... , coefficientN
• (F4: coefficient1, coefficient2 , coefficient3 , ... , coefficientN)

So, imagine that you run an analysis with 12 steps and 4 formants. The file would be 63 lines long. The first three lines would be basic information, then there would be 12 groups of 5 lines, each indicating first the errors and then the regression coefficients for each analysis considered. This information may be useful to consider the winning analysis or alternative analysis quickly and without having to refit regression models.

Editing and Aggregation

processed_data

Processed/aggregated data gets put in this folder. Right now files called 'aggregated_data.csv' (see here for more information) and 'coefficients.csv' are created by default (see here for more information).

formants_edited

Any winning formant files that you put in here will be edited and copies will remain in here. Redundant copies are kept of all files so that any change can be undone and recreated.

images_edited

If you do manually edit any formant tracks, new images will be created for the data and placed in here.