Announcement

Collapse
No announcement yet.

Any matlab users out there?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Any matlab users out there?

    I was recently robbed and among the items taken included a Behringer 8000 microphone that had been calibrated and worked with my DEQX.

    A replacement mic from DEQX will cost 200 dollars--money is tight (having to replace laptops, speakers, TV, much more) and so I'm looking to do a workaround by buying a Dayton clone of that mic whose cal file can be downloaded but which is in a text file format that needs to be translated to a matlab file for DEQX use.

    Any help would be vastly appreciated.
    John S

  • #2
    Re: Any matlab users out there?

    Sorry to hear about that. I have access to Matlab at work and I can do any sort of file conversion that you need.

    Comment


    • #3
      Re: Any matlab users out there?

      Originally posted by Michael P View Post
      Sorry to hear about that. I have access to Matlab at work and I can do any sort of file conversion that you need.
      Very cool of you to offer to help. That will be a huge service.

      I will attach the old mic file to a comment on this thread later and presumably you will be able to open it when you are at work and can see the format.

      I'll buy the new mic today and should have the text file of the mic cal within a few days. I am happy to do whatever I can on this end. Many moons ago I had to use matlab for grad school, and no current access or even excuse good enuf to get a 30 day trial. :(

      Comment


      • #4
        Re: Any matlab users out there?

        Michael,
        PM sent--no way for me to easily attach at these pages as file type unsupported.

        Comment


        • #5
          Re: Any matlab users out there?

          Thanks, I'll take a look this evening.

          I generally avoid Matlab and use Python for most of my work now... I used it in college a lot, but then I found out how much it costs.

          Comment


          • #6
            Re: Any matlab users out there?

            I use MatLab at work, but when I need something at home, I use Octave. It is essentially a freeware clone of MatLab. It runs almost identical to MatLab and most scripts will run without modification.

            Comment


            • #7
              Re: Any matlab users out there?

              Thanks for sending the file. Do you have any graphs of the response that went along with that microphone? I'm asking since I was able to get something looking like a calibration curve out of the file, but even if I guessed the file format correctly, the frequency and amplitude scales could be wrong.

              It looks like the file contains an impulse response. The binary portion of the file is exactly 16384 bytes and contains 4096 32-bit signed integer values. The data looks like this:
              Click image for larger version

Name:	deqx_response_guess.jpg
Views:	1
Size:	62.7 KB
ID:	1161767

              This was done using Python, by the way... I'm not sure there is anything specific to Matlab about the file format (it's definitely not a MAT file).

              Comment


              • #8
                Re: Any matlab users out there?

                Originally posted by Michael P View Post
                Thanks for sending the file. Do you have any graphs of the response that went along with that microphone? I'm asking since I was able to get something looking like a calibration curve out of the file, but even if I guessed the file format correctly, the frequency and amplitude scales could be wrong.

                It looks like the file contains an impulse response. The binary portion of the file is exactly 16384 bytes and contains 4096 32-bit signed integer values. The data looks like this:
                [ATTACH=CONFIG]51279[/ATTACH]

                This was done using Python, by the way... I'm not sure there is anything specific to Matlab about the file format (it's definitely not a MAT file).
                Michael

                Nice work already.Helpful to know that the file is somewhat generic--that was a guess hazarded on the basis that most of the number crunching routines were scripted in Matlab.

                Unfortunately not a thing came with the mic that I recall, but too many years and moves ago to say so for certain. Nor does it at all resemble the curve some chap here had to say about the Behringer mic on a PE review about the FR.

                I am not sure why PE even sells this anymore, since they now offer their own version that comes with its own unique calibration file. Originally the Behringer ecm8000 got a reputation for being reasonably flat at a very cheap price. But for the last couple of years they have changed and the response is more like a typical recording mic with a huge presence boost. I know because I bought one and had it calibrated by a 3rd party (for about $50 extra) . So for example, there was a 5-6 db boost around 8-10 khz. But if you have a good calibration file that is not a big deal.

                But this mic does NOT come with a file and it is not flat. So there is 0 reason to sell the ECM8000 as a "meaurement mic"
                The mic file you posted is from 2005 so it's still entirely possible the review is right and you are also right re the scaling.

                My only use of the product has been as part of the DEQX suite and have never verified the FR independently or used it in an another app. If you do have the scaling correct, then I need to graft/interpolate their (Dayton EMM-6) 256 data point which is comma delimited Hz, +/- dB cal file into a format resembling the DEQX--which may be something like an iFFT. Ultimately my plan is to get a much better mic from Earthworks and for now hopefully get the DEQX to work with the Dayton version. But if it looks like too much trouble I'll suck it up and spend the couple hundred and get the DEQX product.

                In the meantime, let me link this to the omnimic forum and see if folk there with a great deal more experience with mics like these think this looks like a reasonable FR. Once upon a time we had a DEQX specific users group on Yahoo but haven't heard a peep for many moons.

                So again thanks for all the help and good to know about the Matlab alternatives.:applause:
                Last edited by JRS; 10-29-2014, 10:09 AM.

                Comment


                • #9
                  Re: Any matlab users out there?

                  This is extremely speculative, but it's possible they measured the response of the microphone, flipped it upside down (to get a correction filter), restricted it to a limited frequency range (1--20 kHz?) and converted to impulse response for use by the DEQX box. A bump around 10 kHz followed by a roll-off above that sounds reasonable for an ECM8000, though we're missing information about the frequency and amplitude scales.

                  Anyway, I hope you can find out more details about how this works (have you asked DEQX?); you definitely want to be confident that your microphone calibration is working correctly.

                  Comment


                  • #10
                    Re: Any matlab users out there?

                    Originally posted by Michael P View Post
                    This is extremely speculative, but it's possible they measured the response of the microphone, flipped it upside down (to get a correction filter), restricted it to a limited frequency range (1--20 kHz?) and converted to impulse response for use by the DEQX box. A bump around 10 kHz followed by a roll-off above that sounds reasonable for an ECM8000, though we're missing information about the frequency and amplitude scales.

                    Anyway, I hope you can find out more details about how this works (have you asked DEQX?); you definitely want to be confident that your microphone calibration is working correctly.
                    I will work on it--the chief designer/guru at DEQX has always been helpful in the past--not sure how much of this is proprietary--and your surmisal as to implementation seems perfectly reasonable, but I am barely schooled in digital processing. In grad school I found the subject almost as inscrutable as the intro quantum mechanics course I took--principles reasonable; working with the whole complex variable thing is what I never got the hang of--prolly why I chose medicine instead of EE :o

                    What I find interesting is that this many years later, there still isn't a lot of competition in terms of a single box solution to DSP correction of both nearfield speaker amp/phase response as well as DRC (altho I haven't surveyed the offerings personally lately and going on other parties comments) and getting a PC/soundcard to do it reliably and without a whisper remains non-trivial, albeit HUGE progress has been made in this regard. If I had to do it again I might go with the mini-DSP route just because of budgetary restraints. But this much maligned approach (in some circles at least) with its archaic DAC's and 96KHz limits notwithstanding is still able to make my jaw go slack when all is done properly.

                    Curious about what you do in Cambridge? And your own views on digital doctoring....
                    Cheers,
                    John

                    Comment


                    • #11
                      Re: Any matlab users out there?

                      I've used analog active crossovers with limited EQ capabilities for the last several years. I don't have much experience with digital crossovers and EQ, though I think they're definitely the way to go if you can get it right (which it sounds like you have). I plan to switch to digital on my next speaker project, probably a PC-based (or embedded-PC-based) setup with USB DAC[s], or alternatively DSP plate amps.

                      You're right that there aren't a lot of viable "out of the box" systems; I think this is just due to the large amount of effort it takes to design, package, and test everything, as well as the wildly varying preferences and requirements of the (overall fairly small) market.

                      I'm a graduate student in digital circuit design.

                      Comment


                      • #12
                        Re: Any matlab users out there?

                        Michael,

                        Hi. So I took your suggestion and contacted DEQX and they have offered to convert the cal file from Dayton into the proper format for a reasonable fee. I will then have both and we can see the "transfer" fx. I remain curious as to details.

                        As to getting it right, I wish I could take credit but all of that is due DEQX. The software is a bit clunky in spots but if I used it more than every few years, I'm sure in daily use it would be fine. I inhaled deeply and took the plunge into unknown waters by plunking close to 2.5k down back in 2006 for factory refurb and have never regretted it. I used to change out speakers (some of them pretty high end) every few years but this is the one piece of gear that finally solved the black hole of audio for me. YMMV. And I understand for many that passive XO design is a high art and I have nothing but respect for that mix of science/art/experience/and infinite patience it takes to "nail" a design. But for me the digital filters that DEQX output seem superbly transparent, and permit for the combination of a seemingly unlimited number of drivers w/o particular regard for out of pass band nasties or widely differing efficiencies.

                        Anyway, I'll post here or PM when I get some more info. Thanks for sharing your time and expertise.
                        John

                        Comment

                        Working...
                        X