TOPIC:

Let me ask before I start studying the manual and playing with trial version.

Is the FD the right tool to correct a single speaker, a horn? I'd want to correct horn impulse response to get rid HOMs (high order modes). The horn is part of my 4-way system, I use OpenDRC-DA8 for crossing the drivers.

So I could use 2048 taps per channel to correct the mentioned horn which plays 350-10000Hz and does not need steep filtering due to roll off and drivers characteristics.

Hi Mariusz_,

There's no time limit on FIR Designer in demo mode. The final FIR filter export is disabled and only enabled with a license key.

I'd suggest taking multiple measurements within the coverage pattern of the horn, loading them into the "Measurement Averaging" tab, taking the complex average and copying the result to the Import tab, and then proceeding through the regular workflow. There's a general correction tutorial on the website (and also here on the MiniDSP website).

Regarding the "Measurement Averaging" tab:
- Under "Load/Save", use "Select Directory" to choose a directory of measurements. FIRd will attempt to load all valid measurement file types.
- Choose a gain adjustment method for all measurements. The default is an offset of 0dB.
- Under "Reference Measurement," choose a representative measurement either on-axis, or close to on-axis.
-- All other measurements are analysed and time aligned as closely as possible to the Reference measurement
- Under "Reference Measurement," if the IR peak isn't close to time=0 and/or there is high phase slope, click "Find Peak"
-- This time shifts the reference measurement such that the peak is at time=0. All other measurements are shift by the same number of samples.
- Under "Comparison Measurements," go through each of the non-reference measurements and make any necessary time alignment tweaks. Un-check "Use" if the measurement isn't relevant or useful.
-- If you would like to give more priority, in the average, to measurements closer to on-axis, adjust the "Averaging weight" up or down.
- When you're happy with the average, click "Copy Averaged IR to Import Tab."

Kind regards,
Michael

Thank you for hints.

Generally I am looking for the tool to perform the correction like described in the document (see 3.2):

Improving loudspeaker transient

Smoothing the magnitude response and linearising the phase response of a loudspeaker will result in a sharper transient. FIR Designer can do this. Take a look at Fig 12 on the tutorial. The Export tab can display the IR before and after application of the FIR filter.
eclipseaudio.com.au/fir-designer/tutoria...dspeaker-correction/

How successful the filter design is, comes down to how representative the measurements are of the actual horn, and how little the measurements are influenced by the measurement environment. David's paper doesn't go into details about how to obtain suitable measurements. (And I suspect he is using a combination of simulation and measurements.)

At high frequencies, good measurements can be obtained with judicious setup and appropriate windowing/gating. However you mentioned you'd like to correct from as low as 350 Hz. Taking appropriate measurements this low can be tricky.

There are also questions around drive levels since loudspeaker drivers change their characteristics with level. You could try averaging measurements from both different angles (within the horn coverage) and different drive levels. There's no "right" answer and one has to experiment.

I have heard some folks comment that temporally corrected horn speakers sound nice at low to moderate levels, but harsher at high levels. I don't have enough experience with this myself to say.

Kind regards,
Michael

1. I can measure it outside in open air.
2. I have a level which I use almost all the time, and it is rather moderate for long evening listening.
3. My midbass horn can go higher, I can start from 500Hz also.
4. I have made first modelling using my old in the room measurements data. Seems promising.
5. I still have to get idea how to set filter delay and length because in rephase I simply put the lowest possible taps number to get the wanted result.

It's right at the top of the screen: "Filter delay (samples)" and "Filter length (samples)," each with line edit boxes and sliders. Everything, including the final windowed/truncated FIR filter, is calculated and updates in real time. As you make changes, you'll almost instantly see the effects.

Mariusz_ wrote:

5. I still have to get idea how to set filter delay and length because in rephase I simply put the lowest possible taps number to get the wanted result.

I'd highly encourage you to take a look at the tutorials. They are highly detailed, with screen-shots, and after one walkthrough, most folks find the workflow very intuitive.

Kind regards,
Michael