From Computational Memory Lab
Jump to: navigation, search

In order to annotate the session, you need to download some files from the cluster to your machine. To do this, you need the path to the experiment's directory, the subject's id, and the session number, all of which should be listed on the subject database. Once you have that information, you can use to download the data to the correct directory.

There is a Bash script called in the parse_files root directory. You can use this script to automatically download the data for a given session into the correct directory under parse_files. If you invoke it like this:

$ ./

it will ask you to type in the experiment code, subject ID, and session number. You can also specify these arguments directly, like this:

$ ./ catFR LTP031 3

This example will download the .wav and .lst files associated with LTP031's fourth session of catFR. has a number of useful options if you have an unusual setup or are parsing non-LTP data. They are:

-u USER    specify a username for remote host
-h HOST    specify a remote host
-p PATH    specify a remote base path other than /data3/eeg/scalp/ltp

If does not meet your needs, you can still download the files directly using scp. You need to know the experiment code, subject ID, and session directory name for the session you're interested in. You also need the path to the experiment's directory on the cluster; for example, LTP experiments can be found in /data3/eeg/scalp/ltp. The path should be listed on the experiment's spreadsheet. Once you have all that, do the following:

  1. Change to the directory under parse_files of the session you want to work on. This should be under parse_files/[EXPERIMENT NAME]/[SUBJECT NAME]/session_[SESSION NUMBER]. If the directory you need doesn't exist yet, create it with the mkdir command.


$ scp polyn@rhino:/data3/eeg/scalp/ltp/apem_e7_ltp/LTP153/session_4/*.{wav,lst} .

This places the .wav files in the appropriate parse_files directory on your local machine. {{#hierarchy-bottom:}}

For more information on TotalRecall, please click here.