Speech:Spring 2014 Systems Group


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

Groups

 * [Systems Group]
 * Experiment Group
 * Tools Group
 * Data Group
 * Modeling Group


 * Proposal Group

Group Member Logs

 * Arwa Hamdi
 * Valerie Therrien
 * Sinisa Vidic

Assigned machine is: automatix & rome

Undergraduate Research Conference
CIS Capstone: Systems Group

The Systems Group is in charge of any hardware and software changes. Our responsibilities are to maintain and update the current systems infrastructure and system wiki guides implemented by previous system groups. The major task is to run an experiment on Rome and compare it to another run on OpenSuse. Two minor goals are to find a solution to the key generator on Rome(Fedora 19 OS) machine which prevents password-less log in using SSH protocol and perform a backup of the experiments directory. With the help of the Virtual Box, the Systems Group plans to create a virtual simulation of our Caesar and Rome machines in order to find a solution to this bug. The Systems Group will install and format a backup hard drive(300 GB)on Rome for the Experiment Group to use for experiment directory backup.

Weekly Log

 * Week ending 3/18
 * Arwa - Run an experiment & tutorial for installing Fedora on VM.
 * Sinisa - Run a successful experiment, contact Modeling group.
 * Val - Update group log & run an experiment.
 * Group - After 3/19 class, run backups.


 * Week ending 3/4
 * Arwa - Fix hardware page & include Rome, fix Fedora installation issue on VM.
 * Sinisa - Update last log & run new experiment.
 * Val - Abstract proposal for URC, install OpenSuse on VM.
 * Group - After class, more diagnostics on HD.


 * Week ending 2/25
 * Val - Keygen fix and update weekly log, updates on adding Hard Drive in Rome.
 * Arwa - Install Fedora onto VirtualBox and add Rome to hardware page.
 * Sinisa - Ran an experiment (#0190), logged the entry.


 * Week ending 2/18
 * Sinisa - Finished the Systems Group portion of the proposal and uploaded it to the Speech:Spring 2014 Proposal
 * Sinisa - Read information on creating experiment/language model and running train/decoder
 * Sinisa - Researched through Speech wiki for famous Fedora experiment with Error Percentage in 20's, was unable to find it. More info at Sinisa's log under 2/18 section.
 * Arwa - Updated Speech:Hardware page with current disk usage.
 * Val - Added the team schedule to the group log and researched backup ins and outs for tomorrows hard drive installation and formatting for the Experiment Group to use for the experiment directory backup.

Tasks Time Line

 * (Weekly) Val    - Weekly Log update
 * (02/19/14) Group - Add external HD to Rome
 * (2/18/14) Arwa  - Speech:Hardware update
 * (2/18/14) Sinisa - Capstone Proposal Systems Section
 * (2/25/14) Val - Fedora KeyGen issue
 * (3/4/14) All    - Run a train on openSuse OS (Automatix) and Fedora OS (Rome)
 * (3/18/14) All - Write a proposal for or against Fedora OS switch

UPDATED: 2/17/14

To Do's

 * Weekly Group Log updates
 * Speech:Hardware page update with new prices (check ebay, amazon..etc) and any other information needing to be updated. (memory does not have to be upgraded this year)
 * Capstone Spring 2014 Proposal (Sinisa with inputs from Val and Arwa)
 * KeyGen - Need to find a solution to Rome's (Fedora OS) issue with the keygen not working as in openSuse OS (Mike thinks it might be a firewall port setting that we need to change)
 * OS upgrade or transfer to new OS (Fedora) ->This will be the hardest part. We need to run an experiment on our openSuse server (Automatix) and run the same experiment on Fedora server (Rome) and see if it's worth it to move to Fedora from openSuse OS. We need to run a copy of the experiment that David did last year. To make a copy of that file, we need to run this command: cp -r 20149(experiment #). Once we can run it, we need to see what results we get. Last year, that experiment had a 20% error rate. We will need to save the new experiment as a new number (it will be either 149 or 150).


 * We will need to learn how to do experiments/train. http://foss.unh.edu/projects/index.php/Speech:Training
 * We need to find last years experiment that Systems Group ran on Rome. We will use this experiment to use for our testing on both machines.
 * Once we find the experiment we need to make a copy of it since we never use the old experiment for new testing.
 * When we copy the experiment to the new folder we need to update some config files so that they do not conflict with the original experiment
 * We will then need to run 1h train in order to compare the two OS performances of Sphinx
 * Write a proposal for the upgrade or stay with current OS

List of Experiments done using Fedora
 * Speech:Exps 0114
 * Speech:Exps 0115
 * Speech:Exps 0116
 * Speech:Exps 0117

There maybe more of them we will need to look more closely at logs but the above once Eric Beikman did last summer.