I don’t think FF currently supports sweeps from external sources (like for example REW does). What I did was use a toslink from computer to my HiFi for the measurements.
But you can also do the measurements in REW and import them to FF. Never tried this myself but read people who has done it this way. In REW you can save the sweeps as a .wav file and import it into Roon and play it from Roon (but its a little messy to get this working from my experience).
Btw, I would not worry to much about using different streamers. While they can and often do sound different, their measurements will usually be very equal.
You are right with your answer. But what I mean are only the early (or better first) reflections within a timeline of 15 ms after the direct sound arrives to the ears. This reflections should be treated by a minimum of 10 dB. Dirac can do some kind of corrections in this manner. Can FF make this possible too?
I think so, from what I recall from reading about FF earlier they use a moving time window, so maybe start at 10-15 reflections at base and then decreasing that time window as the frequency goes up.
And it seems to working, for example this is the convolution file for left speaker that I currently use. As you can see, there are big and dramatic filters at lower frequency, but as you go towards higher frequencies the corrections becomes more shallow and less steep (lower Q). This is correction strength 5.
Here is the same correction but with correction strength 10, now it corrects more aggressively in higher frequency (which might be good or bad, test and decide for yourself and let the ears decide):
FF don’t contain those kind of “advanced” settings, so not sure how to do that. And for its target audience this is good, nothing scares away new people like a cluttered window with lots of settings. I could do it in REW but than I have to measure again and that’s to much work (I am lazy )
I also have a question, maybe someone in this thread of @Focus_Fidelity can answer.
My step response is shown below, and it looks good except the first ms (red circle). Anyone know the reason for this? Acoustic treatment (I have lots)? Crappy speakers? Interpreting step responses is currently above my paygrade
Just saw this thread and wanted to add my 2 cents. I have used Dirac, REW/Rephase, Serkan’s various methods, HouseCurve, but not Audiolense or Acourate. Focus Fidelity produced much better results and much more easily than anything prior. The bandwidth progression/adjustments allow for full spectrum adjustments (vs limiting to say < 400Hz with some others) and the UI flow is fantastic. I don’t know if I got lucky with most recent measurements/filters but the latest Finetic 2.x version produced even better results. I export convolution filters for HQPlayer and they work beautifully. Well worth the $$.
Looks like tweeter and mid(bass) are originally connected with opposite polarities, as is usually the case with textbook second order passive crossover networks to achieve proper amplitude summation.
The correction inverts the tweeter phase to match the mid(bass) but what remains is a discontinuity in the handover between tweeter and mid(bass) as can be seen by the droop, which should actually show in the amplitude response around the crossover region as well - care to show a measurement of the corrected response?
Unlikely anything to do with room treatments or DAC architecture.
More likely induced by physical speaker and passive network design, as well as desktop setup.
Are your speakers still the Volent VL-2 Paragon?
Looking at (incomplete) measurement graphs here there seems to be an integration “problem” between the drivers in the original design to begin with, which would somewhat help explain your uncorrected step graph.
I think your setup would benefit massively from being freed from the speakers’ passive crossover networks and made fully active.
I just flip the absolute phase to get them to match with uncorrected one. It could be the problem you describe. However, I will test next week to try and get a measurement with up sampling in the mix. Btw, calling out to @jussi_laako which filter would be best to use for measuring purposes? I imagine a very neutral and plain short filter (not that I think it makes a big difference). Or should one measure with the filter used normally?
Here are the amplitudes, faint lines are before correction. Its not a pretty sight, but I have a quite horrible little 3x3m room. Phat base traps means I don’t have any null nodes though so I can correct everything (FF makes sure of this as well).
I’m impressed by the correction provided in the low frequencies with a boost of around 6 to 7 db.
Personally, I’ve only added 2.5 db of boost and that suits me perfectly, above that I think it’s too much.
Well then, there’s the “personal taste / equipment / listening room” correlation.
For measurements, some linear phase filter. Depending on what you are measuring for example poly-sinc-gauss-xl, poly-sinc-gauss-halfband or poly-sinc-gauss-long. (you don’t typically need to correct for errors in the source data)
Well I am a happy camper as well! I purchased this last August and frankly wasn’t satisfied with the results…so went with my usual source—Thierry at Home Audio Fidelity…and still strongly rec that service as I have used it over the past 10 years in four different homes in both my upstairs system and downstairs man cave system. But given all the chatter in this thread and feeling like I had outlaid alot of money just to not use it at all—and knowing there had been some updates, I gave it another go! Last night I still wasn’t happy with the results—seemed a bit panned to the right…so redid the filter today with the original measurements and left the very step regarding auto zeroing alot and voila…perfect phantom center and a wide and deep soundstage! Very satisfied with results! Using it with HqPlayer all at DSD 256 and with some updates in HqPlayer (cannot rec highly enough and used it since 2012) can run my fav filter, modulator, and the FF convolution filter without any pauses or dropouts and it sounds great! Give it a try!
Doug
@Douglasmaurer, I didn’t understand the following sentence. I’m interested in learning how you achieved perfect center balance and deeper soundstage.
…redid the filter today with the original measurements and left the very step regarding auto zeroing alot and voila…perfect phantom center and a wide and deep soundstage!
When I first did the filter I followed the directions a bit too literally regarding choosing the primary listening measurement and auto zeroing the impulse measurements. The manual is a bit unclear on whether this step is necessary. I did it the first time and must have messed it up and the center image seemed a bit off to the right. So i reimported the measurements and simply accepted the default measurement zero positions and skipped to the next step and this time the filter generated was perfectly centered.
I went back into the filter creation steps with Focus Fidelity and saw that my impulse measurements were auto zeroed as well. I reset them and uploaded the convolution filter into Roon which fixed my issues - the same as what you described. Thanks for sharing and helping me out in the process!
)
