Pyult

Minor bug fixed.

From this release, you can select only one frame from each recording to be converted into dataframes and images. For this, you need to place a configuration csv file named "frames.csv" in a directory (preferably named "config") in the working directory. This file should have two columns, the first one of which should be a stem name of each recording and the other of which is the frame to be picked. So, for example, if you have "xxxUS.txt", then the stem name of this recording is "xxx".

The last version couldn't handle with the images where a tongue surface curve is rather obvious (i.e. when only one peak can be detected for each scanline). This bug has been fixed in this present version.

The last version gave an error and stopped working when it encounters erroneous images, which cannot be fitted with spline curves, such as completely black images. This problem is fixed. The current version (0.0.4) simply returns input images and fill output dataframes with the y_spline column filled by NaN.

Fixed the bug, which occurred when attempting to produce dataframes from images with fitted splines.

Removed unnecessary newline markers at the ends of words and segments.
Also, the __main__.py now automatically removes noise rows, e.g. \<P\>, based on words and segments columns.