Yeah, I understand that, clearly caused by Roon switching the indicator at the 100x threshold. It would be better if it wouldn’t do that 
More concerned with the result not matching the prediction, I must have messed up something
Not as if you look at it all the time though is it well I hope not.
Sure, a minor thing, but still
Roons full of minor things that never go away unfortunately.
3 Likes
Well yeah, I held the iPad all wrong 
New pictures tomorrow. Just to be sure, when the app says “top iPad microphone”, this is what it means, yes?
Yep, that’s right The app will say which one it is using (it picks the default for the device).
It is normal for measurements to vary with location. I recommend taking a few measurements in your listening area and using the average for the correction. To compare to the prediction, take the same measurements again and compare the average. It should be pretty close. A single measurement is harder because you must measure from exactly the same location (use a stand, etc).
Yes it says top, just confused what top is New to iPads, no idea what the default is. All clear now, thanks
Right, good point. I averaged for the correction, but not for the comparison. I’m sure I’ll be less clumsy tomorrow. Anyway, it did good things even so
Tks Greg, your input is much appreciated.
Another question: Is there a downside to using more filters? The manual says
For a convolution engine (ie: using the impulse response), the number of filters just controls the fidelity of the correction. More filters may not make an audible difference.
But more filters than the default 10 do even out some more individual peaks. So I wonder why not use more?
I think, the manual wants to point out the difference between
a) using the EQ Settings calculated by House Curve by either manually transferring them to the DSP device or use some export file format
b) creating convolution filters inside House Curve and import those to your digital audio player
With a) you are limited by the capabilities or the specific device to which you are transferring the EQ settings in terms of max. number of filters, max. dB correction etc.
This is usually the case, as the playback device needs to do all calculations on the fly in real-time (or faster) during playback and therefore the limits are dictated by hardware capabilities.
With b), all individual EQ corrections are put together into one single impulse response and it doesn’t matter, how many are being used and how complex they are as the playback device is only presented with one mathematical operation (convolution) which is almost trivial for a stream of discrete values like PCM coded music.
So the CPU power needed for this is alway the same, regardless of the number of filters „hiding“ inside the convolution file. It‘s only influenced by sample rate, bit depth and the length (I.e. number of samples) of the convolution file.
This is why convolution takes a lot less cpu than PEQ which can easily be observed with the signal path speed indicator.
After reading again, I suppose your question is about the „more filters may not make an audible difference“ statement.
I think @GregD would be able to give a conclusive answer to this, but from my understanding, House Curve is designed to distribute the number of available filters according to the severity of corrections needed.
I.e. first the big peaks and nulls, then the smaller ones. So increasing the number of filters beyond a certain point only adds very small additional corrections, which might not be audible in the end.
Yes, you‘re right. Autocompletion strikes again 
1 Like
Need some help. Bought HC and fail to measure anything. BT connected to a Primare NP5 streamer and later to a Buchardt I150 directly by AirPlay. Using an iPhone 12pro and then an iPad Air. Both on iOS 16.3. For both the microphone is allowed with HC and indicated during measurement. In every combination I cannot create a single successful measurement. Both pink noise and sweep fail. The test tones are audible though I cannot identify the initial chirp. I get ‚listening for test signal’ then the sweep starts (can hear no chirp before) the mic/recording indicator is on and later HC times out and shows ‚Could not detect test signal’. Any ideas welcome.
Thanks for both of your answers, but indeed the latter. Sure, the corrections become smaller with higher filter numbers, but one seemingly not wholly insignificant peak in my curve is flattened when i go from 15 to 16 filters. So if there is no downside, why not use them. What you wrote for case (b) in your first answer would be the reason why I would crank up the filter number - if there is nothing else that comes into play.
Maybe the chirp gets lost on connection initialization to the playback device?
You could try to export the measurement sweep to a file and play it back through roon.
Check „stimulus type“ in the measurement setup dialogue of HouseCurve for this.
This way, you can also do measurements with the corrections in place to check on the results afterwards.
1 Like
I would see the number of 10 filters as a rule of thumb.
The point from where more filters don‘t make much of a difference obviously depends on the individual situation, in your case it seems to be 16.
I think the only possible downside of using more filters „than necessary“ might be with additional phase shifts created by those filters that might have a negative impact on performance in terms of clarity and spatial presentation. However, a small correction like 1 or 2 dB usually doesn’t create large phase shifts either.
2 Likes
Will try right after siesta 
here. Thx
Next to the suggestion of @Roland_von_Unruh, I also made the external sweep a repeating playlist of 1 track, so I can walk around the room making measurements… I start a test and when I fail to include the chirp (long walk to the next test location…) I have to wait for the next chirp otherwise I get the same error message as you’re mentioning…
Good luck!
2 Likes