Speech:Spring 2017 John Vey Log


 * Home
 * Semesters
 * Spring 2017
 * Proposal
 * Report
 * Information - General Project Information
 * Experiments - List of speech experiments

Week Ending February 7th, 2017
2/2: Set-up Project binder for the Data Team documentation. I will take charge and make sure the group documentation from logs etc is kept in order and will bring to class each week. Binder currently includesSpeech:Spring 2016 Data Group instructions as well as the Data Team's porposal from last semester. Begin to disect all former Data Team's information with a very sharp scalpel. Ongoing.
 * Task:

2/3: Set-up SLACK to communicate with team. Tried to log in to pubpages from Chrome to do my logs and and got a blank page. Hit pubpages and the Internet with a shoe until I could log in.

Start preliminary work to present to the team as far as the proposal goes. See below.

As of 2:26 PM, read through all student logs.

2/6: Created Helpful Links on Main Page for Data Team. Full disclosure: Dylan found them, I just posted them to the main page, but I have encouraged the team to add as necessary. Created links for Videolan and Filezilla.

Communicated with team via Slack today, addressing concerns about experiments and proposals.

2/7: Reviewing work of previous teams in order to come up with some sort of prototype timeline.

Created prototpye timeline - needs considerable team input. Posting to Slack.

1:32 PM: Read through all of my team members logs.

2/2: No results at this time.
 * Results:

2/3: Got SLACK working for the team and class and was able to log in.

2/6: No results at this time.

2/7: Posted to Slack 1:27PM. Waiting for responses from rest of team.

2/2: I assure you, you may wonder why the heck we need a physical project binder. If it's not on paper, it didn't happen. It's a much better visual representation of progress and will be much faster to check on our work then constantly looking things up on line, plus we can keep track of project annotation as we go along. When we need it, you will all be glad we have one.
 * Plan:

Preliminary review of last semester's Data Team leaves me with several strong concerns especially in regards to their past work and their recommendations going forward. Taking notes and coming up with plan for proposal.

2/3: Looking at last semesters work, it seems that the last semester's data team was working towards a ~40% accuracy goal as far as the data. Their recommended goal for our semester is to shoot for 60%. Please see Concerns as to why I disagree drastically with this idea, while at the same time I'm still not sure what the plan should be going forward. I have some ideas, but I'll need team input.

2/7: Timeline: We're winging it because we have no guidence other that what others have done in the past. While I usually would not be so open as to using what came before as a template (we don't know if someone elses work got them a C or worse), if this is research aren't we expected to build on other work?

2/2: Only concerns with the project binder, as always, is that it's kept up to date. If it all goes to hell with the binder, it's my responsibilty. I'm not worried about my group, but it's also a good way of tracking team members who don't tow the line as far as contributing to the project. I highly recommend every team have one.
 * Concerns:

VERY concerned with what I've read so far in regards to last semester's team work. Not to be critical but I've spent years doing work in regards to data gathering and analysis in both research and industrial settings. Trying to come up with reasonable goals to present to the team.

2/3: Having worked with data anaylsis, especially as far as quality control issues are concerned, I do NOT recommend going with last semester's recommendation. It is flawed, but I understand what they were trying to do. Their recommendation is to shoot for ~60% accuracy. Now, I understand that quality improvements in a given system give exponential improvements, not linear. While I quickly agree that the 20% jump in quality data would in fact provide a considerable jump in data quality, in regards to the entire research project over-all going forward the next several years of semester, it's far from good enough. Look at it this way - 60% is a huge jump, but it's D-grade work. Imaging the data was a student. Do you want to be working on a Capstone project and get your data from a student who's getting a D? No! You want at least a C, maybe a C+. Bare minimum, we should be shooting for at least ~70%-75% so that the rest of the teams are not working with the results of "junk data". Long term, future semesters should be doing a "moonshot" of at least 80%-85%.

Small improvements to the quality data will result in exponential improvements to the entire research project.

It's not fair to the other teams to do work with poor data, when we have a chance to really push things and improve them.

Read through all student logs. Concerned about lack of logs.

2/6: Concerned I'm going to be an idiot again and forget to save my previews like I did 2/4 and already once today. Concerned about group use of outside messaging systems as in I don't see consistent use yet. Still early in the project so I'm not very worried yet. I don't want "We only meet once a week" to become an excuse (work from home is the trend - not the exception).

Concerned about the Draft Proposal. While it says the assignment is due Wednesday, it also says that class discussion will be the proposal and the timeline. Is it due beginning of class or end of class? Are we expected to work as a group then or before class starts. Either way I have discussed with some of the group about cobbling together something before class starts with limited guidence.

Concerned that experiments are being run without group input or agreement. However, we need to get moving. Doing something is still better than doing nothing.

2/7: Timeline prototype: Unfortunately we have no guidence or guideline as to how this is supposed to be done so I only hope what we come up with is acceptable. Concerned about team logs.

Week Ending February 14, 2017
2/8: Class prep - reviewed all logs from team. Updated project documentation in project book. Included all logs as of 2/7 and proposal/timeline update. Added Prototype timeline (template off of 2014 timeline for data team). Worked togoether to finalize draft proposal
 * Task:

2/12: Working on final proposal - ongoing. Read over all of my teams logs. Communications over Slack: Posted Dylan's page on how we can improve the dictionary from Sorceforge CMU Sphinx Page.

2/14: Checking logs and messages. Going over documentation for class tomorrow.

2/8: Good news! Vitali from the Modeling Group confirmed that the data from last semester is a mess. It's not just our teams inexperience; the data really is a mess.
 * Results:

2/8: Going forward, we have data to try to help straighten up for modeling; more data to gather and a final propsal to work on.
 * Plan:

2/8: Had to hash out confusing terminology (mixed terminology) left behind from last Spring semester.
 * Concerns:

2/12: Concerned why after setting it up and testing it, Slack isn't sending me instant messages from team. Concerned when team is finding issues with the project, they make corrections, figure things out but don't communicate to the rest of the team what they did. We had agreed on writing up series of tutorials for what we did, building on the work of others first.

2/14: Concerned I had a lot of other things going on this week and was not able to give the attention I wanted to to the class. Fortunately I'm wrapping up the last of this business project and I'm done that for the semester. I'll be able to do some catch up this weekend.

Week Ending February 21, 2017
2/15: Class- MJ and Dylan out with flu and I'm behind. Will be taking group notes and scanning them for the rest of the team.
 * Task:

2/18: No notes to scan for team. Co-ordinated with Matt who co-ordinated w/ Prof. Jonas over the proposal. Copy of new draft proposal saved, review and printed out for Master Documentation Book. Reviewed Slack communications. Sent out Slack message INSISTING we must me online consistantly. Sent message out to Shay (she's going to help with this tech that I just can't comprehend and no one else will help me with - at least she understands/accomodates my disabilites.) Realized there is a setting on the main page of Slack that sends out messages to "@everybody" that "Everybody" on my team wasn't getting.

Read through team logs. Read through ALL Slack messages for ALL teams.

2/19: Log into wiki and get to work. Review team logs. Print out all of Brenden Collins logs from 2016 (reference recommended by team members) for reference - some of us still need things printed on paper.

Watched videos of Sphinx (https://www.youtube.com/watch?v=IAHH6-t9jK0).

Read 'RESEGMENTATION OF SWITCHBOARD' white paper by Neeraj Deshmukh, Aravind Ganapathiraju, Andi Gleeson, Jonathan Hamaker, Joseph Picone (http://groups.inf.ed.ac.uk/switchboard/reseg-swbd.pdf) as recommended by Brenden Collins.

Finally figured out connection issues from home. Logged into caesar.unh.edu.

2/21: Preparation for class. Reading through Tucker's slack comments so we all have uniformity of voice.

2/18: I'm only hearing back from Matt, but he's coming back with results. I agree with Matt that Matt is going to handle communicating the proposal to the rest of the class, everyone else will need to contirbute something and I have the time to make it "one voice" per requirements.
 * Results:

2/19: Made significant progress. Learning what I need to figure out. Shout out to Shay for helping! RESEGMENTATION OF SWITCHBOARD whitepaper short, but very helpful.

2/15: Class - Status report the best I can with limited team. Take notes for everyone and send slack messages.
 * Plan:

2/18: Keep moving forward.

2/19: Plan is to put in full day of 8+ hours work to catch up. (EDIT: Passed 8 hours an hour ago).

2/18: I'm concerned there will be resistance/excuses for not meeting online once a week. Concerned overall about levels of slack communications HOWEVER - after looking through the other teams Slack communications, we're doing better than some of the other teams.
 * Concerns:

2/19: Missing instructions in wiki did not help connection issues. Rusty with Linux/Unix.

Week Ending February 28, 2017

 * Task:

2/22: Finalizing proposal online with team. Reviewing Plans and setting tasks for individuals with team.

2/26: Plowing through the text files as per instructions. See Matt's log entries for parameters.

2/27: Same as above.

2/28: Same as above but read through a considerable log post and documentation posting from Dylan.


 * Results:


 * Plan:

2/22: Post and Fianlize Proposal. Going well.

2/26: All goes according to plan. Not waiting on script.
 * Concerns:

2/22: We're posting our Final Proposal today. After last week I'm concerned about everything, especailly the unexpected.

2/27: Concerned we're going to get way ahead of things (that's not such a bad problem to have).

Week Ending March 7, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending March 21, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending March 28, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 4, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 11, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 18, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 25, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending May 2, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending May 9, 2017

 * Task:


 * Results:


 * Plan:


 * Concerns: