Friday, February 10, 2017

Week 1

Hello everyone. It's been a week since my project officially started, and I am glad to say that everything is going well. This past week, my goal was to find a direction and define a schedule for the rest of these 10 weeks. I was able to see the lab for the first time, and meet a few of the awesome people who work in it.

The lab is part of ASUs Department of Speech and Hearing, found in the glass Coor Hall. Here Dr. Berisha introduced me to the students working with him on his many different projects. Afterwards Dr. Berisha gave me many options to define a more specific direction for my research, so I decided it would be best to meet with each student in the lab and learn more about their projects. 


Lattie F. Coor Hall

The first student I met with was Prad Kadambi. His project, a little newer than the others, was based on the cough. A new device is being developed to be worn by a patient, so that every time they cough the device records the type and amount of coughs. Prad is helping develop the algorithms and code for this device. 

The next day, I met with the remaining three students in the lab: Ming Tu, Yishan Jiao, and Alan Wisler. Their research is a little further along than the cough project, so there are many opportunities for me to learn from what they have achieved and still watch how they work. Ming has done work with speech from patients with Dysarthria, or impaired speech. He was looking at how computers can improve the assessment of the severity of Dysarthria. His work has also spread to the realm of automatic speech detection(e.g. Siri) . Yishan's work looks at the different features of speech and how they can be analyzed. She has worked on analyzing rhythm (through speaking rate), articulation, and pitch, with which she can use to learn many characteristics of the speaker. Finally, I met with Alan. His research dealt with a lot of the theory behind each algorithm of the other projects, relating them to the field of machine learning. He works with probability theory and statistics to find ways to improve upon algorithms. 

So... what's the plan? With so many options available to me, like helping develop a new product, looking at previous works to see their processes, or learning about the theory of it all, I struggled to pick just one project. With that, I decided to spend the next few weeks learning even more about each project by looking at each individually. After that, I will see if I can apply what I have learned, from the theory and other projects, to the the cough project. 

Now that I have this plan and support from Dr. Berisha to make sure I learn the most in these next nine weeks, I am even more excited to see what I can learn. I will share as much as I can with you guys, so stay tuned and feel free to ask any questions. 

Until next week...Buh Bye.


30 comments:

  1. Alright, second week and things are going well I see. You mentioned that you would like to explore a bit more of what their projects have to offer, and is there any of them that you are most excited to learn more about? Personally I would choose the one where you could have a tangible product. I think that is really cool!

    ReplyDelete
    Replies
    1. Right now I'm really excited to learn about all the techniques they use in programming these algorithms. But the project that is the most exciting for me is the cough project. This is because there are more opportunities to program something myself and make an impact on a physical piece of hardware.

      Delete
  2. Hi Luke! It looks like you've learned a lot about Speech research and the cough project in your first week. So every week will you be focusing on a new project? Also, will your research be mostly focusing on the algorithms behind each project?

    ReplyDelete
    Replies
    1. For the first few weeks I will be looking at all the projects at the same time (if that is even possible) to see if I can learn enough to focus and make an impact on the cough project.Yeah, so most of the features and characteristics of speech are analyzed by using the algorithms, so by focusing on them, hopefully I can pick up some knowledge. If i do that I will be able to shift my focus from learning the algorithms to applying them and looking at speech samples.

      Delete
  3. Hey Luke! It seems like you're getting the chance to learn about so many different aspects of speech and hearing at your site! Which of the four project are you most interesting to learn more about? Also, how are planning on implementing what you learn there in your own project?

    ReplyDelete
    Replies
    1. I am really interested in seeing how the research that is further along has accomplished the results they have, which seem pretty promising. As for my own project, I will probably be merging my project with the cough project, so the final product of mine will hopefully be a part of the cough device.

      Delete
  4. Hi Luke! It's nice to know that you have a plan going forward, it makes things a lot easier for the coming weeks! Is this what you were expecting out of your senior project? Did you expect there to be so many relations between statistics and speech patterns?

    ReplyDelete
    Replies
    1. When I was first looking for a project and found this one, I was not expecting this much statistics and probability. But it makes complete sense why there would be so much relation when trying to pick out patterns and features from samples of speech with so much data and information.

      Delete
  5. Herro Luke! I'm glad that you were able to establish a plan on the first week. The expansion of your senior research project into multiple ideas is very intriguing -- I was just wondering, are you planning on implementing all the ideas into your senior research project or finding one idea that's suitable for your project?

    ReplyDelete
    Replies
    1. Hey Evan, my plan for implementing these ideas centers around the cough project. My goal is to be able to incorporate the ideas I learn about into the cough project to see if I can make an impact.

      Delete
  6. Wow, that's a lot of info. Some questions though -- do you know how exactly they will measure the severity of dysarthria? Will it be qualitative (Mild, Severe, Extreme, etc) or quantitative (like on a scale from 1-10000000000000)?

    ReplyDelete
    Replies
    1. The generally accepted measurement of dysarthria is quantitative on a scale from 1 to 7, with 7 being the most severe or atypical and 1 being typical. To get to this scale doctors or listeners look for different features like articulation and slurring, they then rate those to get the overall rating. The problem is, this can be very subjective because some listeners put more importance on different features.

      Delete
  7. Hey Luke! These projects you have been learning more about sound really interesting! You mentioned linking the research's algorithms to the "field of machine learning." What exactly is this field? Also, what does it mean for the furthering of the speech projects? Looking forward for next week's update!

    ReplyDelete
    Replies
    1. The field of machine learning in this case relates to the training of the algorithms. Instead of making algorithms that have the importance of features and different factors hard coded into them (which would be extremely hard), they are trying to make algorithms that can be trained on sample data to decide the importance and factors on its own. This means a more efficient way of making even more accurate algorithms(assuming they are trained on a lot of good samples).

      Delete
  8. Hi Luke, I'm glad that your first week has gone so well. Of the four students, which project did you think was the most interesting and that you would want to learn more about?

    ReplyDelete
    Replies
    1. I am excited to learn about all four of them, but for now I am most interested in the projects further along. I want to learn about how the projects with strong results, like Yishan's and Ming's, have accomplished what they have. I am also interested in learning more about Alan's theory, because it helps explain how everything works.

      Delete
  9. Hi Luke! Your first week has gone very well and I'm excited for your project as it progresses. The method you used to approach the initial stages of this project seem effective and I like how you are going to first learn where you stand before developing an idea fully. What inspired you to choose this topic in the first place, and what pointed you towards this lab?

    ReplyDelete
    Replies
    1. Hey Adi, when I was looking for a project I really wanted to find something that would mix the arts or music with physics. When looking through research in the physics department at ASU, I came across Dr. Berisha's work. When I got here, he was very open to letting me look at all of his projects, leading me to the plan I have now.

      Delete
  10. Hello Luke. It's nice to see that you've made progress on your project and developed a plan. Is there a specific reason why you're choosing to apply your learning to the cough project, or is it just because it's the newest? I think all of these projects are really interesting and I look forward to seeing what you learn from each.

    ReplyDelete
    Replies
    1. Thanks Jeffrey. Because the cough project is the newest I thought I could make a larger impact on the project with my limited/new knowledge.

      Delete
  11. Hi Luke. You mentioned automatic speech detection and its relation to this research. I wonder if you could use some knowledge about these existing programs, for example, Siri, as you said, to jump start the project. In other words, what can you learn from the progress of such a popular system in terms of detecting diseases through speech?

    ReplyDelete
    Replies
    1. Something that is a goal for improving Siri is being able to adjust how Siri speaks to the way the user speaks. This means they are trying to look for features of the users speech and define the characteristics of the user. This is very similar to detecting diseases through speech because it uses similar techniques to look at features of speech.

      Delete
  12. Hey Luke! This cough project sounds very interesting. What kinds of algorithms/ patterns are you expecting to see throughout this experiment? Can't wait to check in next week!

    ReplyDelete
    Replies
    1. One thing that everyone in the lab was mentioning was Deep Neural networks. They were saying that is was a pretty popular way of looking at data. I will talk more about this in the next week update, so look out for that.

      Delete
  13. This comment has been removed by the author.

    ReplyDelete
  14. Hi Luke! What a great opportunity that you will be able to see so many projects at different points in their research. In regards to the cough device, how will this device be used in the future? Will recording the type and number of coughs help doctors and researchers determine the type of illness someone is possibly suffering from?

    ReplyDelete
    Replies
    1. Exactly, Dr. Berisha explained it as similar to the fitbit, where a person wears the device and it tells them information about their health possible illnesses. I am not sure how wide spread this device will be, or if it will be mainly a tool for doctors.

      Delete
  15. Looking forward to hearing more in the next few weeks. But I am glad you had the opportunity to discuss a variety of options with faculty and students in the group. Some invaluable first hand lab experience. Enjoy the process.

    ReplyDelete
    Replies
    1. Thanks Dr. Sahu, I am getting a lot of exposure to so many interesting projects. I am trying to make the most of it, and I'll keep you updated.

      Delete
  16. Hi Luke! I'm sorry for commenting a little late, but i'll make sure to keep up and not forget. I think its interesting that we can use sound to predict our health. But does the sound predict severity of a disease or a disease itself? Or both?

    ReplyDelete