WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions.
Click here to have your audio equipment measured for free!
I don't think that will do what you think it will. It simply adjusts the sample values in the convolution file, scaling them to the peak value to help use more of the bits for better resolution.
But, you can see the approximate headroom you need when you select the impulse response file in the convolution screen in HQPlayer. This is a good thing to check when selecting files, just to be safe.
Thanks again. So there's no way in REW to just have this headroom "baked in" into the impulse response WAV file?
A simple way to reduce all the PEQ band gains by one number for example? Would be nice to do it on that "EQ Filters" screen, because that's where the 'headroom required' is showing too.
I use HQP Embedded and don't see any headroom info . So I'm trying to make the WAV files 'idiot proof' so in HQPlayer I only need to leave headroom at -3dBFS and never have to change.
I got the below reply from Mr. Harman Curve himself below.
And by the way, I haven't given advice to anyone in this thread, other than to tweak to personal taste...
I emphasised a few times that I'll probably end up liking something that's in between Harman Curve (well Oratory1990's EQ to be correct) and flat line, and so far this is proving to be the case.
Actually, I’m with you on this. For me, Harman target sounds OK, but not great. I EQ my phones to a nearly a flat line with in-ear mics and then bump up 3-4dB below 200Hz to taste. Maybe it’s just what I’m used to, but this sounds just much more natural to me. I’ve not found too many here that agree with my method, though, I have to warn you
One consideration is that I do listen a lot to binaurally recorded music. A flat target curve seems to makes a difference in this case, since some of the pinna effects are already baked into the recording. I still do like "flat" curves, even on non-binaural recordings. But that may just be my personal foible.
One consideration is that I do listen a lot to binaurally recorded music. A flat target curve seems to makes a difference in this case, since some of the pinna effects are already baked into the recording. I still do like "flat" curves, even on non-binaural recordings. But that may just be my personal foible.
I mean, if we measure a headphone with a curve very similar to the harman curve with a pinna measurement system, it should result in a reference curve that we can use as we measure withouth pinna recording system. is it so?
Here's a step by step guide on how I do it. It's a compromise between the general unreliability of the measurements below 60Hz and above 6000Hz as well as the limitations of different PEQ software: APO can only do 0.5dB gain steps, qudelix can't do Q above 7, rephase has poor accuracy for low frequencies, etc. First you get measurements of your headphones in this format, lines starting with * can be removed and the phase column is optional. Do some replacements if you got the measurements in a different format, pay attention to dots and commas.
Also take note of what fixture was used to take those measurements - if it's not standard then target curves won't be of much use. Autoeq's github has some raw measurements or you can use network activity of your browser to get them from various online graphing tools. Then you acquire a target curve in the same manner. Try different ones to see which you like more. You can even try using different headphones as a target, that's arguably even more reliable(as long as they were measured on the same fixture). Then you set target curve as the house curve and use these settings for eq - the default ones are set up for speakers.
Then you press file -> import frequency response, then press the big EQ button in the main window, hide the waterfall graph by pressing the triangles above it, set the logarithmic view and set graph boundaries using the buttons on top and adjust the target level so everything is visible and set flatness target to 1dB. You may limit max boosts but I just set them to maximum values Then you adjust the target level such that subbass wouldn't need equalizing, set that as the lower bound of the match range, then you find where the target intersects with the measurements in the upper treble and set that as the upper bound.
From here you either start equalizing yourself or press the "match target" button(ignore the warning) - it will already produce a decent EQ, if a bit excessive in the number of filters.
Then you adjust them to your taste, try different targets, adjust for limitations of your EQ software, whatever. The goal is to make them sound right to YOU after all.
If you need convolution I strongly advise to enter the filters you designed in rephase instead, there's way more stuff you can do there, but more caveats as well.
Just click on the EQ Filters button at the top, and enter the desired filter parameters. When finished, go to File->Export->Filters impulse response as WAV. Select Mono, Current Filters, 32-bit Float, and the desired sampling rate(s). Save the WAV file then use it in your convolver. Here's what it looks like for the settings you quoted from Amir: View attachment 107085