Speech:Spring 2013 Charles Haynes Log


 * Home
 * Semesters
 * Spring 2013
 * Proposal
 * Report

Week Ending February 5th, 2013

 * Task: Figure out what this class is all about. Try and grasp a basic understanding of what speech recognition is, and what tools are are required to performs experiments.


 * Results: I think I am more confused then when I started. There seems to be a large lexicon associated with speech recognition I have never been exposed to.


 * Plan: Read over the ALL the sections on the wiki and dive in head first


 * Concerns:I am really foggy at this point on the direction of where the class is headed.

Week Ending February 12, 2013

 * Task: Read and compare previous semesters of goals and contributions of the Speech Project


 * Results: Good, I have a much better understanding of the types of programs our group will be working with, and what we need to do to progress forward.


 * Plan: Finish group proposal listing the goals for the next 4 weeks.


 * Concerns: None at this time.

Week Ending February 19, 2013

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending February 26, 2013

 * Task: Work with Lee to help figure out how to run a train using CMU Sphinx tools.


 * Results: Ran into difficulty using Linux due to Sphinx being command line based, and us being unfamiliar with such commands.


 * Plan: Continue to plow ahead with a 2nd computer open to Google for questions.


 * Concerns: Wondering if my Linux skills will ever be up to par...

Week Ending March 5, 2013

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending March 12, 2013

 * Task: Create a virtual machine with documentation for upcoming semesters. The tools group has discovered having a separate machine as a testing environment would be very helpful in testing upgrades to the software package.


 * Results: Making progress towards establishing a stable running environment. The machine is not fully operational yet, but progress has been made with documentation being built along the way.


 * Plan: Continue forward


 * Concerns: Creating proper documentation has been a daunting task, do to the trial and error approach being taken to create the machine. I am afraid there will be holes in the document.

Week Ending March 26, 2013

 * Task: Rally the group the find out what they have been up to for the past couple weeks. Tasks were divided and we need to get together to find out what direction we are headed in.  Attempt to complete the virtual machine testing environment and start planning for the Capstone poster.


 * Results: Group is doing well with Bego spearheading Sphinx knowledge. It was determined an upgrade to the CMU Sphinx tools was not needed this semester.  This is due in part of Prof Jonas not wanted to upgrade to Sphinx 4 from 3, as Sphinx 4 is java based.  Some Sphinx forums have whispers of a major update happening soon.


 * Plan: Finalize the the Capstone poster layout and create the PowerPoint slide from Microsoft WebApps so the group can work on it in real time.


 * Concerns: What is left to do in the tools group besides the completion of the virtual machine?

Week Ending April 2, 2013

 * Task: Start an email chain that includes all members of our new Modeling groups. Tap the resource of my classmates who have a much better understanding of how the speech train and model are produced and tested.  Figure out what GenTrans is all about and what changes we can make to it.


 * Results: Followed the email chain to find out what other group members are up to and went through Tommy and Eric logs understand what modeling the data.


 * Plan: Meet with the group to help run a train all the way through to scoring.


 * Concerns: Matt is working on the GenTrans to try and improve the accuracy of the model, but I do not how the changes are making will make a difference.

Week Ending April 9, 2013

 * Task: Run a train with the help of the group on batch machine Majestix. Touch base with Group A to check progress on rebuilt corpus for new 5 hour train.  Attempt to run a 5 hour train through to completion and check word error rates.  Read updated logs from classmates.


 * Results: Completed train 0072 all the way through to SCLite with the help of group D. Talked with fellow group members about the status of the 5 hour train and what I could contribute to making progress.  Found out the corpus was built, completed, and ready to test.


 * Plan: Touch based with group leader to help figure out what time of day works best for the group to attempt the five hour train together. Continue Gmail conversation with group members to ensure we are all on the same page moving forward.


 * Concerns: A five hour train is going to take forever...

Week Ending April 16, 2013

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 23, 2013

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 30, 2013

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending May 7, 2013

 * Task: Collaborate with group members to help coordinate Final Paper. Contribute to the final paper introduction and conclusion section.


 * Results: Went very well. The group contributed to almost fully complete the final paper.


 * Plan: Have a great summer and start to a career


 * Concerns:None it's finally over!