Originally Posted by jrett
Any chance we could get this as an option in the UI? I'm currently researching how to do this with the import function (which I can't find in the manual fwiw) and I don't want any house curve other than this one.
Specifying target curves is an area where MSO needs improvement. I'll put out an interim solution in the next version, and for the version after that I'll try to come up with a better way.
I just recently got some new information from
about the target curve from the Harman experiments as a result of a discussion in another thread
. He posted a link to a PDF file
of the Harman study that shows they used a second-order shelving filter to create the target curve. This shelving filter had a "half boost frequency" (sometimes called "center frequency") of 105 Hz, which was not changed in the study. Only the boost was changed.
This filter type is the same as MSO's second-order low-frequency shelf. However, MSO specifies such filters in terms of of the boost and the low-frequency corner. If you keep the low-frequency corner constant and vary the boost, the center frequency will change. Part of the interim solution for target curves will consist of defining an alternate form for first- and second-order low-frequency shelving filters which allows the user to specify the boost and the center frequency (instead of the boost and the low-frequency corner). The other part of the interim solution will be an option to save a filter channel response as a target curve. So the interim method for the next version would consist of the following:
- Create a dummy configuration that is completely empty except for a single dummy subwoofer channel.
- Add a shelving filter of the alternate form as described above to the dummy filter channel, allowing the specification of its boost and center frequency.
- Make a graph showing the response of this dummy subwoofer channel.
- Tweak the filter parameters to taste.
- A new command will be implemented for the filter channel allowing you to save the filter channel's response as a target curve text file.
This is clearly a kluge, and what's needed is a new method dedicated to only specifying target curves, without the need for dummy configurations and so on. I'll do that in the version after next. That version will also add cubic spline interpolation to help in specifying smooth target curves with a small number of points.
By the way, Geddes recommends a target curve that rises at 3-6 dB per decade below 200 Hz. This can be easily specified in a simple target curve file as follows (example is 6 dB boost). For this example, "Use logarithmic interpolation" should be checked.
At frequencies lower than the lowest frequency specified in the target curve, the target value will be the same as that of the lowest frequency, so it will level out at 6 dB below 20 Hz in the example above. A similar thing happens at the high-frequency end of the curve.