4

I'm currently working on multiple mini pet-projects in my spare time, grounding myself in the basics of audio signal processing, and trying to develop algorithms for polyphonic pitch detection. As I'm having a lot of joy (even in the failure) in doing so, I am considering moving into a career in signal processing (I'm aware this is a vast area!) My background is in mathematics (I have a bachelors degree in applied mathematics, so the theory and practice excites me) and I have no grounding whatsoever in electrical or electronic engineering.

I'm therefore wondering the following:

  1. Assuming I can develop a project portfolio demonstrating basic competence, what specific careers would be open to someone of my background in something highly geared to signal processing? I would be particularly interested in sound engineering, but would also be interested in compression algorithms and coding theory.
  2. Is a supplementary MSc. necessary to get my foot in the door to job positions? I am open to both industry and academic positions.
  3. Would an MSc. in something involving machine learning and neural networks be a.) relevant, and b.) worth the dough? I am considering getting out of industry for a year to work more on my projects in the context of such a degree programme. I am getting the impression that combined with a knowledge of signal processing, some knowledge of A.I. and self-learning systems would go a long way.

Thanks in advance for your answers.

robert bristow-johnson
  • 20,661
  • 4
  • 38
  • 76
Andrew Whelan
  • 143
  • 4

3 Answers3

2

In general, Maths to audio signal processing is a very natural progression. I seem to recall my Master's and PhD supervisors both did their undergraduate degrees in Maths.

To your questions:

  1. This is the hardest question to answer. Depends what's in your "portfolio of basic competence". In addition to the understanding of signal processing theory that you're developing you need to show you're able to implement these algorithms probably either in hardware or software. Given little EE grounding, I'd suggest software. Look at developing your software engineering skills. They will serve you well in either academia or industry.

  2. MSc will definitely help but is not absolutely necessary. The types of algorithms you are describing could conceivably be the subject of an MSc level project. I work at a company that makes hardware for audio signal processing. The statistic they gave me is that around 60% of staff have a Master's or higher degree.

  3. Yes, combining signal processing and machine learning is a very wise idea if you want to work towards the research end of the field. Several problems in audio signal processing - e.g. audio source separation and speech recognition- are increasingly finding their state-of-the-art solutions come from machine learning.

tobassist
  • 816
  • 5
  • 18
1

You can always turn a good mathematician into a good electrical engineer, but not necessarily with the converse. In my opinion.

that said, if i were you, i would consider, if you do go back to school, taking a good course in DSP that uses basic Linear System Theory but goes beyond it.

probably, since your interest seems to be in music and audio (whatever else is a polyphonic pitch detector good for?), i would consider getting an MS or whatever from the CCRMA program at Stanford or CNMAT at Berkeley. Or maybe something at U Miami, but i dunno exactly what department.

if you're not going to work on hardware (nor even spec the hardware), then i dunno what good it is to know much more from Electrical Engineering other than the mathematical discipline you get from Linear System Theory (sometimes called "Signals and Systems", O&W have the standard text on that), Communications Systems, Control Systems, and Digital Signal Processing. A course in Electronics might be helpful to give you an idea what Analog signal processing might be, but you might never ever need that if what you'll be doing is solely writing math and writing code to define and implement algorithms.

robert bristow-johnson
  • 20,661
  • 4
  • 38
  • 76
1

The problem with asking your question here is that you trigger a self validation response in most people who are inclined to answer. Seeing someone show interest in what you do “feels good” and it validates the life choices you made. No body is going to say that if they had to do it over, they would be an accountant. If I were born to a wealthy family, I can’t say that I would choose DSP over painting water colors of nudes in the south of France.

DSP is not a bad career choice. If you put something into it, you will get something out.

Somethings that make something bad is what makes it good. You can have a wide range interests but something like reading an article in IEEE Signal Processing magazine can reveal how little you know, Some people think this is good to have challenges, others see Sisyphus

If you want a job, just start applying for those jobs. Most employers are looking to extract 60+ hours of work out of you for the least pay, which is often good pay.

AI is a hot topic. At my school, the AI related classes are bursting with people. It is exciting but scrubbing data is what is mostly what people do.