http://www.twilley.org/morse/index.htmlI put some time and effort into building a software package that generates new practice files every night and makes them available via RSS. I've tested it for the past few days and it seems stable so I thought I'd share it here. If you're interested in learning Morse code, please give this method a try. Let me know if you have any suggestions on improving the software. It works with iTunes, Google Reader, and Thunderbird so it should work with most other podcast browsers.The MP3 files have three parts: a header, where the new character is repeated with code and voice five times; a body, containing ten lines of ten five-letter groups each; and a footer, where the answer key for the body is spelled out. This removes the need for a text file attachment for the answer key and means folks can practice anywhere. The software uses the Koch method to select the characters for practice and both standard and Farnsworth spacing. The voice is mine -- I'm going to try to get other people with different accents and dictions to provide audio files so when I release the software, there will be options for those who wish to hear another voice.On a slightly related note, I've been grappling with a minor defect with the Koch method as originally defined. The standard for moving on to the next letter is 95% accuracy or higher, yet data files contain an even distribution of all known characters. Once a student gets past 20 characters, the new character is statistically likely to occur less than 5% of the time, so it would be possible to "pass" without actually knowing the new character. The final round of test has fifty unique characters, which means it is probably only going to occur 10 times in each 500-character set, nowhere near often enough to demonstrate proficiency with that character. My idea on how to deal with this is to generate test files with only the ten most recently learned characters (as an example, the thirteenth set of practice files which would normally have the characters KMRSUAPTLOWI.N would only have UAPTLOWI.N) so each recent character occurs often enough to impact the final score. The downside is the potential loss of memory of the early characters, but hopefully by the time practice gets that far they'll be burned into the student's brains. What do you all think?