aj94GXE

I guess for anyone who's into modding their car stereos this is slightly on topic. This semester, as my term project for Speech Recognition Class, I created a full Natural Language Processing system for Music Selection, and since I couldn't demo it in a car, I made it work on Winamp (a rather common media player). It works so nicely that I really want to put it in my max (and with 3 grand I could afford to get all the hardware to do that).
You guys can see a quick video clip of it working here:

http://users.snip.net/~juneja/speech/heymrdj.avi

and the presentation here:

http://users.snip.net/~juneja/speech/heymrdj.pdf


You'll need the DivX codec to view it... www.divx.com has that. Note that this video has no sound, because when capturing, if I tried capturing audio or increasing the frame rate any, it would be too demanding to run the speech recognition engine.

Anyway... this leads me to an interesting dilemma as to what to do with this summer's internship (FYI, I'm a junior at Carnegie Mellon in Pittsburgh doing a Double major in Mech E. and CS and a minor in robotics).

Here are my options, and I'd like to hear from fellow car enthusiasts and college students what they would do in my situation -- these are all places I have pending interviews for January and February:

1. The Bosch Research and Technology Center in Pittsburgh. For all of you that don't know, Bosch is a very prominent German parts supplier that happened to make the first spark plug and the first Anti-Lock braking system. They are working on things such as speech systems for automobiles, and if I'm there long enough I might get to make my own project up because Jill, the person whom I talked to there "could tell I'm a person who likes to dream up projects..." (how right she is

Note that this lab is very new and currently only has 6 people in it. I met them at the job fair at CMU in September... and my conversation with Jill ended with her telling me to "definately follow up in December." to which I replied, "Oh, in that case I'll send you a Christmas card!"

Well, December came and went (Happy new year everybody!) and the day after my birthday (Dec 6th) I went over to there office to surprise them with a Christmas card and some cookies. I planned to be there for five minutes... but ended up staying there an hour and a half after saying "yes" to Jill's offer of coffee even though I really didn't want any coffee. I talked to two of the engineers and Jill during that time about various things from work related stuff to dancing, and I left thinking that I basically had an informal interview there... and boy were they nice to me!

Pay rate: $25-30 an hour, or about 12K pretax for the summer.

Also, a friend of mine, Chris, a masters student in Electrical & Computer Engineering, interned there last summer, and got to test his wireless network for the car in a BMW M5! Lucky bastard...

2. IBM -- I was surprised by this one, but given that I worked for a different dept. out in California last year, I'm not too surprised that the Pittsburgh office called me. The two groups I will be interviewing with in January are "NextWeb" and "WebSphere Edge Server."

WebSphere Edge Server is a caching server that is quite a cool product, if I do say so myself... and I'll get to get into lots of nitty gritty cs stuff with it, so that's kind of cool.

NextWeb is a division that works closely with IBM research to create the future internet technologies... things for pervasive computing... and some of this stuff would overlap slightly with what Bosch is doing with Wireless Ad-Hoc Networks in cars... more stuff I find really cool.

As a slight pressure, a certain Distinguished Researcher at IBM (my mentor) would really like to see me back there for this coming summer. A couple other people high up in IBM would like to as well.

I was supposed to interview during finals week, but I called to push it back because my speech project died that monday and needed to be brought back to life while I studied for finals. So I will interview there in 3 weeks.

Pay rate: $23 an hour. IBM pays based on the number of credits I have... so no bargaining here.


3. At the beginning of this semester, I proposed an accent filtering idea to help foreigners speak better English (and not screw up voice recognition engines so much). For example, native Chinese speakers often pronounce an "l" at the end of the word as an "r."

So the idea of this is to analyze the inconsistencies in the waveforms and reshape them so that they are "more correct" all while keeping the person's own voice.

Anyway so I started this project this semester, and haven't gotten very far yet... because of my coursework taking priority, and the professor I proposed this to, Maxine Eskinazi in the Language Technologies Institute (LTI) at CMU *really* wants me to devote the summer to it, and publish a paper, and get a patent... and she really wants a feature like this to go into her product, Native Accent, which is a computer based program to train people to speak better English.

I'm reluctant to commit because it will only pay $3000 for the summer although possibly more when a National Science Foundation grant kicks in in August. Though this project is sort of my own creation... and seeing it work would be wonderful...

Maxine has gone so far as to start bribing me... she gave me a scarf for Christmas, invited me to a ski trip at her condo.

So... all of these opportunities are great... but how do I decide? Which one would you guys and gals choose? Everyone at CMU says publishing a paper is really valuable for graduate school, but last I checked, so is working at research labs at very large well known companies such as IBM and Bosch! And each of these paths ultimately leads down a different path in life for me... as getting full time offers at the end of this summer is a possibility!


Thanks for reading all this, and when I have the funds, I may make my maxima into a testbed for all the speech stuff for the car stereo...

Sprint

iTrader: (10)

too long.. you got cliff notes?

ngthing

your divix is mirrored...can you fix that?

as for summer...

i'd do something you'd like to do...the paper/patent or something like that, don't worry about the money..obviosuly you're a bright person, that'll all fall in to place later. as for me...i'll be hanging out at our state univeristy

aj94GXE

I don't think the divx is mirrored... no one's told me that yet. I once had that problem... do you have version 4.02 or newer of the codec? I used 4.02 to make the file, and hypercam to capture it.

As for what to do... I'm not quite sure what I would like most right now. I'm kind of afraid that I don't know enough speech recognition stuff yet to pull off the accent filtering thing properly. I need to take a DSP course which I can't seem to fit in anywhere, since it's in Electrical Engineering at my school.

aj94GXE

1. Made a nifty speech system for a car stereo. Look at the video and report links above.

2. Gotta decide what to do this summer between:

1. Bosch Research Labs -- things like speech systems for cars.
Pros: lots of money, cool projects, cool cars, and beautiful women.
Cons: Beautiful women. that's a distraction to have at work. Especially when one said person is only 6 months older than me, and still single. (I'm 21). I don't think I mentioned this, but said person is also extrodinarily beautiful... and really friendly, too.

2. IBM -- things like pervasive computing and web caching.
Pros: decent amt. of money, cool projects, and they called me! Already worked for an IBM dept. before, though.
Cons: already worked at an IBM dept. before, mentor pressure to return to IBM...

3. Accent filtering project -- my own little dreamed up creation.
Pros: was my idea to do this in August.
Cons: now the professor likes the idea, and wants me to commit fully to it. unfortunately it pays peanuts.

ngthing

ahh it was the codec. looks neat. i added you tomy icq list, maybe i will catch you online.

mrb23

maybe she's bribing you cuz she wants you, not just the project. :P

cool work though, go for what you think will help you most in the future