Speech:Spring 2014 Valerie Therrien Log


 * Home
 * Semesters
 * Spring 2014
 * Proposal
 * Report
 * Information - General Project Information
 * Experiments - List of speech experiments

Week Ending February 4th, 2014

 * Task:
 * Read previous Systems group logs
 * Access and log into Caesar (change password)
 * Look up latest version of OpenSuse
 * Research and learn more Linux commands

2/4
 * Results:


 * Read logs
 * Started practicing and using Linux commands at work

2/5


 * Read system group logs from 2013
 * Tried to access the Caesar server: caesar.unh.edu - couldn't access it, will ask my team members.
 * Researched on the OpenSuse site over the weekend for the latest version. Looks like the latest version is 13.1. The previous year upgraded it to version 12 but according to the OpenSuse website, they are no longer supporting that version. I think we should definitely do something about that! It also looks like v13.1 takes about 4GB of space. The newest upgrade is faster in maintaining performance when it's doing disk operations. It has also been out for more than a year, so most of the bugs have been fixed already.
 * Searched for more complex linux commands that may be helpful.
 * Concerns:
 * If we upgrade to OpenSuse 13.1 will we have enough space on the servers?
 * When can we do backups?
 * Should we upgrade to OpenSuse or Fedora?

Week Ending February 11, 2014
2/5 2/11
 * Task:
 * Meeting at 1pm with Systems group and Mike - take notes and add them to our Systems page
 * Find instructions on how to start the experiments/training
 * Talk with team members and figure out our specific roles
 * Results:
 * Updated our systems page with notes from the meeting
 * Read logs
 * Found experiment instructions in the Summer 2011 Training
 * My roles: Weekly updates on System group log & KeyGen task


 * Plan:


 * Concerns:

Week Ending February 18, 2014

 * Task:
 * Update System Group page
 * Add HD to Rome (02/19/14)
 * Read Sinisa's Proposal & check for changes
 * Backups after class (02/19/14)
 * Find out commands on how to format hard-drives
 * Learn how to ssh to rome

02/17/14 cd /dev -> point to the directory ls hd* -> this will list the hard drives in the server sudo fdisk /dev/sdb -> to clear the partitions sudo mount /dev/sda1 /where/you/want/the/drive/to/be/mounted
 * Results:
 * Added our group schedule to the System's page
 * Read logs and Sinisa's proposal - didn't make any changes
 * Read group emails about fixing errors in the proposal
 * On 02/19/14 at 4pm our group and the Experiment group will take the 300 GB hard drive from Mike and we will need to format it (using ntfs) and then put it into Rome. Then the experiment group will backup all of the existing experiments on the machine onto that hard drive.
 * Some of the commands to format the hard drive:

Helpful link: http://www.runtime-era.com/2012/01/format-external-hard-drive-using.html


 * Concerns:
 * Are we fully in charge of backups?
 * Will we be continuing to use Clonezilla like the previous years?

Week Ending February 25, 2014

 * Task:
 * Install Hard Drive on Rome with Systems group
 * CMOS battery issue in Rome
 * Fix KeyGen issue
 * Weekly log for Systems group

2/20 We viewed the hard drives in Rome by:
 * Results:
 * Rome was not labelled at all. We first had trouble finding out which one it was, we eventually found Rome's IP address (192.168.10.11) and followed the network wire that was labelled 10.11. Rome is the Dimension 8400. We found that Rome is the Dell Dimension 8400 (underneath the Verleihnix server). To get Rome out, we had to turn off and unplug Verleihnix (with the permission of the Experiment group) and pull it forward to be able to take Rome out. Rome can only get taken out by the front of the rack because of all the wires in the back of the servers. Once we got Rome out, we noticed that there was only one more bay for a second hard drive. Rome only takes SATA wire connections for the hard drives. After checking all of the utility closets, Jonas was the one who had the correct wires. Then we installed the hard drive and put Rome back. Then we powered on Verleihnix (was down for about 30 minutes). When we booted Rome back up, we got an error saying: "System Battery Voltage is Low". To fix this, we need to get another CMOS battery for the motherboard. They are inexpensive but tricky to find. I will research and find one that we can use.
 * Log into Rome as Root
 * sudo fdisk -l
 * parted \dev\sda2 print (this lists the drives available in the system)

We tried a couple more commands to add the new hard drive for the storage in Rome but it doesn't see the drive at all. We will be meeting next week to take Rome apart again and make sure the connections are in the right place and if that doesn't work, we will go into the BIOS and make sure that Drive 2 is enabled! Right now, Rome has a secondary drive but it doesn't see it.

2/25
 * CMOS battery needed is model#: CR2032. I found one that I can bring in next week!
 * Need to look at the Firewall settings on Rome for KeyGen fix

Week Ending March 4, 2014

 * Task:
 * Write abstract for URC
 * Learn how to start an experiment - ask Sinisa
 * Need to install OpenSuse & compare

3/4
 * Results:
 * Abstract is on the Systems group page
 * Fedora is installed on my VM, but still need to get OpenSuse installed

3/5 Will need to get permission from Experiments group to take down Verleihnix for about 30 minutes while we work on Rome.
 * Meet with Systems group after class and work on Rome (Hard Drive issues).

ssh-keygen -t rsa
 * Check firewall port settings: "start the firewall-config tool and select the network zone whose settings you want to change. Select the Ports tab and the click the Add button on the right hand side. When the Port and Protocol window opens, enter the port number or range of ports to permit.

Week Ending March 18, 2014

 * Task:
 * Ask our group for some more advice on Keygen issue.
 * Start first experiment but talk to Sinisa first before starting.

3/15
 * Results:
 * Spoke with our group to discuss next week's goals. We will be all meeting after the Wednesday class to do our first backup of all the experiments.
 * Sinisa told us what experiment to start with. Will start and record my first experiment on Sunday.

3/16
 * Read logs & found steps on doing the experiment.

3/17
 * Update System's group page
 * Created first experiment, will be documenting issues & results later.

3/18
 * Process of running first experiment, Speech:Training page has been very helpful. When I went to make my first experiment, I went into /mnt/main/Exp # mkdir 0214, but then got this error message: cannot create directory `0214': File exists. There is no log on the Speech:Exps page about experiment 0214, so I tried 0218 which was the next one & that experiment got the same error. My experiment is now 0219. I am doing a duplicate of Exp 0163 as my first experiment. I ran % /mnt/main/root/tools/SphinxTrain-1.0/scripts_pl/setup_SphinxTrain.pl -task  inside my experiment directory and then it said: Generating SphinxTrain specific scripts and config file, Set up for acoustic training for 0219 complete. This looks like good progress. For the setting up of the Sphinx Train configuration file, this website is very helpful in learning how to insert/text in the command prompt: http://www.adminschoice.com/vi-editor-quick-reference/. After I edited the lines 6 through 8, I got this message:


 * 1) removed from the filler dictionary and added to the dictionary (this
 * 2) is because the force alignment is not very good at inserting them)


 * 1) $CFG_FORCE_ALIGN_DICTIONARY = "$ST::CFG_BASE_DIR/falignout$ST::CFG_EXPTNAME.falign.dict";;
 * 2) $CFG_FORCE_ALIGN_FILLERDICT = "$ST::CFG_BASE_DIR/falignout/$ST::CFG_EXPTNAME.falign.fdict";;

$CFG_FORCE_ALIGN_BEAM = 1e-60;
 * 1) Use a particular beam width for force alignment.  The wider
 * 2) (i.e. smaller numerically) the beam, the fewer sentences will be
 * 3) rejected for bad alignment.

$CFG_LDA_MLLT = 'no'; $CFG_LDA_DIMENSION = 29;
 * 1) Calculate an LDA/MLLT transform?
 * 1) Dimensionality of LDA/MLLT output

$CFG_CONVERGENCE_RATIO = 0.04;
 * 1) set convergence_ratio = 0.004

$CFG_QUEUE_TYPE = "Queue";
 * 1) Queue::POSIX for multiple CPUs on a local machine
 * 2) Queue::PBS to use a PBS/TORQUE queue

$CFG_QUEUE_NAME = "workq";
 * 1) Name of queue to use for PBS/TORQUE

$CFG_MAKE_QUESTS = "yes"; $CFG_QUESTION_SET = "${CFG_BASE_DIR}/model_architecture/${CFG_EXPTNAME}.tree_questions";
 * 1) (yes/no) Build questions for decision tree clustering automatically
 * 1) If CFG_MAKE_QUESTS is yes, questions are written to this file.
 * 2) If CFG_MAKE_QUESTS is no, questions are read from this file.
 * 1) $CFG_QUESTION_SET = "${CFG_BASE_DIR}/linguistic_questions";

$CFG_CP_OPERATION = "${CFG_BASE_DIR}/model_architecture/${CFG_EXPTNAME}.cpmeanvar";

$CFG_DONE = 1; When I press Esc, all I get is "Return 1;", I'm not sure what this means?? I'm stuck on this message, my next step is to generate the transcript message. I'm going to ask my group & the class if I can't figure this out today.
 * 1) This variable has to be defined, otherwise utils.pl will not load.

Week Ending March 25, 2014

 * Task:


 * Run more experiments and read logs

3/24
 * Results:
 * Read logs

3/25
 * Successfully ran an experiment following the Master Script.
 * Will need to work on Decoding. I haven't done much of that yet.. will talk to group members to see if they can help me out.
 * Will be meeting with System's group to discuss un-finished projects (keygen) and backups that need to be completed. We will be meeting after class is over to go into server room and fix a couple issues on Rome. We will need to have Verhlenix(Experiment group's server) down for about 30 minutes while we are working on taking Rome out of the rack.

Week Ending April 1, 2014

 * Task:
 * Do final System's Group proposal
 * Work on System's Group draft of Poster
 * Help Sinisa with doing KeyGen work?... Jonas has 2 different ideas of finding the solution.
 * Need to talk to Experiments group next week and show them where to start saving copies of the experiments.
 * Need to label Rome on the KVM switch in the server room. It's not label correctly right now!

2/26 These are the commands to format and make a directory to save the files to:
 * Results:
 * We first ran Clonezilla (the disk is on top of the spare Win 7 computer in the server room) on Rome to make a copy of Rome's hard drive, and followed the steps using the Backups page on the Information page. We were first going to make a copy of Rome on the new 300GB hard drive that we had installed a couple weeks ago but Jonas told us that we didn't need an extra backup of Rome. So, we reformatted the 300GB hard drive again.

* mkfs -t ext4 /dev/sdb1

* mke2fs -t ext /dev/sdb1

* mkfs.ext4 /dev/sdb1

* mkdir Exp_Backup This creates a new directory called "Exp_Backup"

* mount /dev/dsb1/Exp_Backup This mounts the new directory "Exp_Backup"

* rm - r test2345345 This removes the test folder "test2345345" we made before creating the "Exp_Backup" folder

2/27
 * Read logs and worked on System's group final proposal.
 * Read URC website to start making the poster for the System's Group.

2/30
 * Read more logs & started doing System Group's final proposal
 * Started designing our poster for URC, but I want to check out last year's poster design so I don't make it completely different!

4/1
 * System's page has been updated with final proposal.
 * Posted rough draft of URC System's group poster.

Week Ending April 8, 2014

 * Task:
 * Change backup directory to mnt/backup instead on Caesar so everyone can access that directory
 * Look into Automatix -> Basically it boots up and unlike obelix, gives network access. Gnome doesn't start and hangs forever
 * Help Justice League group with running a 100 hour train

4/7
 * Results:
 * Read logs & online forms for creating directories / deleting and renaming directories

4/8 Delete specific command: rm -r test234535folder
 * Created the backup directory under Caesar's mnt. Its under: mnt/backup (this is where we will now save the experiments)
 * When I went to check out the "Exp_Backup" directory that we previously made, I didn't find it. Either someone renamed it to be just "backup" or "Exp". I checked under Exp & there are 0253 experiments saved in there. We will need to copy these files to the new directory in Caesar: /mnt/backup.

Week Ending April 15, 2014
4/14
 * Task:
 * Worked on poster & read logs

4/15
 * Read group member's logs
 * Final poster has been completed, will be emailing it to Prof Jonas.
 * Working on the final proposal with Arwa, we will be updating it on the System's Group page.

Week Ending April 22, 2014

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 29, 2014

 * Task:
 * Finish System's Group Final proposal for Fedora.
 * Run Last_5hr train and report results to group.

4/26
 * Results:
 * Logged in, read logs
 * Reviewed current Fedora proposal
 * Will be running train on Sunday

Week Ending May 6, 2014

 * Task:


 * Results:


 * Plan:


 * Concerns: