Topic: Sony ECM MS957 - amateur frequency response chart

[naturenutt]

NOTE:  In the following test, I did NOT account for the frequency coloration of the speaker that produced the sound that the Sony ECM MS957 mic recorded.  I don't consider the frequency chart I posted here very accurate or usable.  Please refer to my most recent post/test on 10/12/2018 further down for a more accurate test and frequency response chart.


I've searched for several years to find a frequency response chart for an old stereo mic I have, a Sony ECM MS957.  However, I've never found one.  So, I finally decided to create my own.  It's not anything professional, just an amateur's approach - no anechoic chamber, just a well-damped room from many books, bookshelves and boxes.  I setup and aimed the mic at 1 meter (approx. 3 feet and 3") from a full-range monitor speaker (20-20kHz), which I've used for mixing for several years.  I ran a linear frequency sweep to the speaker from my computer, then recorded it with the MS957 mic into a Marantz PMD661 as a 24-bit/96kHz wav file.  I used software to process the generated freq sweep and the mic recording of it, coming up with the chart below.  I mainly did this because I wanted to flatten out several nature recordings I'd made over the years with this mic to see if would help them.  It helped a lot - smoothed them out quite nicely.  Hopefully of use to anyone else looking for info on this old stereo mic.

Chart can also be found here on flickr:

https://flic.kr/p/DVGohq

[DSatz]

Nice project.

IF you wanted to go further with this type of thing (and that's a big "if", I realize)--here are some things to consider. You measured on-axis response in a quasi-anechoic environment from a moderately close distance (1 meter). At 1 meter, you can have pretty good control over room reflections at mid and high frequencies--but not at low frequencies. Also, proximity effect is still a factor, boosting the (measured but not real, in your application) low frequencies even further.

So I get the impression that this microphone's response was designed for an overall impression of "clarity" rather than neutrality (= typical of Sony consumer/"prosumer" electrets for decades). I think you would be justified in correcting the low frequencies via an even greater boost than the one your curve suggests. (Your curve doesn't indicate the difference between "M" and "S" channel response, but usually the "S" channel's low frequencies can stand even more of a boost, if you're equalizing them separately.)

--Another suggestion is to measure the response at (say) 45° off-axis, and overlay the results onto your 0° curve to give you some impression of the microphone's diffuse-field response. This can be quite different from the on-axis response, particularly at both ends of the spectrum. When you're recording in a three-dimensional sound environment, the majority of your signal arrives from well off-axis, so you'd want to consider that as well when deciding on a corrective curve.

Finally, I don't know whether your software provides options for smoothing or not, but if so, don't be afraid to use some amount of it. The "jaggies" (large apparent differences in sensitivity at nearly adjacent frequencies) aren't generally meaningful; they could be due to resonances in your microphone, but more often they're due to other artifacts of the measurement setup. If the mike had been four inches closer to the sound source or four inches farther away, the jaggies you'd get would be largely different. Another possible way to reduce them is to use third-octave noise bands instead of a frequency sweep.

--best regards

[PaulCayard]

I used software to process the generated freq sweep and the mic recording of it

Which software did you use to extract data? Is it free? Thanks for further explanation :-)

[naturenutt]

@DSatz:  I really like your ideas for further testing of the mic.  I do record both far-field and near-field nature stuff with this mic.  Just for fun, I'm going to try more testing of this mic (Sony Ecm Ms957) using your suggestions.  Your ideas got me to thinking that I'd like to re-test with a matrix of these approaches:  on-axis and off-axis, close and far, mid & side, and left & right.  Then, I want to create different combinations of overlays of the results, which I'll post back to here, along with the individual results.  I'll also see what I can do in the software for smoothing the responses graphically - I'm still getting the hang of the software (trial) I'm using.

@PaulCayard:  The software I used to extract data is a program called SpectraPLUS (I'm currently experimenting with a 30-day trial of it, which is fully functional).  It's not free, but out of several software programs I've tried (free and trial alike), it has the easiest-to-read spectrum I've looked at so far.  I'm still searching for a free one that's as easy to read, preferably with various coloration options of the plotted frequencies.   The SpectraPLUS software does have real-time analysis and offline analysis.  I've been using the offline analysis of the .wav files that I made when testing the Sony mic.
..............

I've recorded a lot of sounds over time with this particular mic.  For examples of those sounds, you can find those recordings on www.freesound.org - just search on 'naturenutt'.  Some of those sounds are with the Sony Ecm Ms957, and some are with 2 Rode NT1A's.  I mention which mic/s I've used in the detail write-up of each recording.

[PaulCayard]

@PaulCayard:  The software I used to extract data is a program called SpectraPLUS (I'm currently experimenting with a 30-day trial of it, which is fully functional).

Thank you very much for this information. It seems perfect for my purpose :-)

EDIT: I've succesfully used it! Result on http://taperssection.com/index.php?topic=176617.msg2179875#msg2179875

[naturenutt]

Finally getting back to this post after 2 1/2 years.  I recently conducted a pink noise test with the Sony ECM-MS957 stereo mic.  I had 3 things to overcome.  One, I had no idea what the frequency curve was of my custom built speaker(s); Two, I needed some baselines; Three, I had other mics with published frequency response curves, so that really helped, but I had to setup frequency response curves to flatten their sound first.  I'm unsure how accurate my 1st frequency response post was 2 1/2 years ago.  It did not account for frequency coloration in the speaker I used for both the test then and the one now.  I did account for it in this new test.

So, the unknown I wanted was the frequency response curve of the Sony ECM-MS957 microphone.  My knowns were a pink noise file and the frequency response of the Rode NT1A mic.  So, I used the NT1A and the pink noise file to first determine the frequency response curve of my custom speaker.  From that I was able to create an EQ curve that would flatten the pink noise coming out of the speaker.  Once this once done, I was able to test the Sony ECM-MS957 stereo mic with fairly flat pink noise audio.  My results are as follows in the attached graphic of the frequency response of the Sony ECM-MS957 stereo mic.  I also explained all of my testing in the graphic, including info about the speaker used, positioning, pink noise volume, audio file normalization, etc.  One thing I did not touch on in the graphic is why I chose -16.3 dB LUFS for normalization of the source and recorded pink noise files.  Normally, I would have aimed for approx. -14 dB LUFS (music-related) or -20 dB LUFS (movie-related).  For my purposes, I typically aim for around -16 dB when processing nature recordings, so that's roughly what I aimed for when I normalized the pink noise source file and recorded microphone file.

Hopefully this amateur chart is of help/use to anyone searching for a frequency response chart of the Sony ECM-MS957 stereo mic.

Chart can also be found here on flicr:

https://www.flickr.com/photos/139094646@N08/43538612680/in/dateposted-public/