esbrewer - try doing the sweep again, though you should be able to see
anything odd by comparing the sweeps in Audacity.
BTW - I have found a way of graphing the freq responses *with* mic
calibration. Attached are the same graphs as before (blue original
measured, light blue Minimal RC measured,
mortod;232798 Wrote:
BTW - I have found a way of graphing the freq responses *with* mic
calibration.
Can you please provide more info on how you did this?
Thanks,
muski
--
muski
SB3-Bryston BP25DA-Bryston 4B SST-Wilson Watt Puppy 7
Transporter via XLR-Headroom Max Balanced Headphone
Isn't your final filter, for example normal.wav, exactly that, with mic
correction and applied filter?
--
nuhi
nuhi's Profile: http://forums.slimdevices.com/member.php?userid=10571
View this thread:
I was plotting the NoCorrection TestConvolution, but looking again I'm
not so sure that that is the same as the measured impulse response with
mic correction (you would expect the NoCorrection filter to be a flat
line but it is not). So what is the purpose of the NoCorrection filter?
--
mortod
My understanding is that the NoCorrection filter does freq response
correction only (ie no phase correction). Also, I believe that the
TestConvolution files don't apply mic correction.
An earlier post mentioned that the REQW developers might add a feature
to apply a mic cal file to an imported
Just noticed (after finally doing some listening after all this grunt
work) that the HF response in the right channel only is way down - to
the point where I thought I had possibly blown a tweeter running the
sweeps.
When set DRC to None - the HF comes right back to the right speaker.
Perhaps
inguz;229908 Wrote:
No, sweep with EQ runs the whole processing pipeline. I think you may
be hitting up against the inaccuracy of EQ bands at low frequencies
(where, essentially, the EQ filters aren't long enough for the slopes
to reach the EQ strength you want). I have other people
Nice to see this is working. I'd be curious to know how you think the
17-band EQ sounds compared to Minimum. It looks like Minimum provides a
somewhat flatter response across the mid-band.
--
tonyptony
tonyptony's
You're right - just listening to David Gilmour, his voice sounds fuller
with correction, I guess because some of those lower frequencies are
relatively higher.
--
mortod
mortod's Profile:
Isn't this stuff great? :-o
I remember years ago being so unhappy that I couldn't afford an Audio
Control C-101 equalizer. Times sure have changed!
--
tonyptony
tonyptony's Profile:
Sure is. Another one for the wish list. Is there anything that can be
done about cpu usage? One thought, could we have the option of
streaming in WAV over the wire(less) (rather than converting back to
flac)? I assume that would save some cpu.
Are there any more light weight tools that we could
I ran new measurements without the sub in my system, and the large dip
around 130hz went away. The crossover must have been doing something
odd - thanks for pointing me in the right direction guys.
But DRC is still not acting on the upward trend in the HF. Attached
are my new measurements
esbrewer;232504 Wrote:
But DRC is still not acting on the upward trend in the HF. Attached are
my new measurements (measured response, minimal, normal, insane). Any
suggestions as to why?
I suggest reviewing the target curve file (PSPointsFile) and mic
calibration file (MCPointsFile).
esbrewer;231946 Wrote:
I checked the pspoints file, and it is indeed set to the bk-3-spline
target. I have no explanation as to why DRC is allowing the HF
increase to remain. It is my understanding that the HFs are rolled off
to varying degrees in the bk target files. I've been reading
tonyptony;232034 Wrote:
Hmm, possible. I assume you're using an ecm8000 mic correction file?
I wonder if you take a new Sweep w/EQ capture, using one of the
corrections that you've shown us here, whether you'd get a similar
actual response.
I believe all of my .drc files refer to the
tonyptony;231155 Wrote:
Hmm, that's the Normal test convolution? In the Inguzaudio\Tools
directory find the normal.drc file and open it in Notepad or similar
editor. Look for PSPointsFile; see what's after the equals sign. If
it's bk-3-spline.txt then I'm not quite sure why it isn't
mortod;230718 Wrote:
DRC applies mic calibration to the filter, not the TestConvolution which
means that all those frequency responses are without mic calibration and
*are* comparable.
Thank you -- an explanation that makes sense! Now I understand why my
normal flat Test Convolutions look
Something looks odd at 150Hz - perhaps a crossover for a sub? You're
asking RC to do a lot to correct that. Normally a poor room leads to
resonance (ie higher peaks) at certain low frequencies. DRC would have
to add a lot of gain at 150Hz to correct that which would cause a lot
of clipping. I
Hmm, that's the Normal test convolution? In the Inguzaudio\Tools
directory find the normal.drc file and open it in Notepad or similar
editor. Look for PSPointsFile; see what's after the equals sign. If
it's bk-3-spline.txt then I'm not quite sure why it isn't correcting
the slope of the high end
tonyptony wrote:
I agree with
mortod that you could probably do better by trying to correct that dip
with moving the speakers around or maybe trying some room treatment.
Or maybe moving the microphone/listening position.
___
audiophiles mailing
Thanks for the input.
My Paradigm Studio 20s run off the high pass out of a BW ASW600 sub.
It sounds reasonable that the crossover may be causing that nasty dip
at 130, and contributing to the clipping problem. I guess I'm asking
DRC to make too large a correction in that area.
The Studio 20s
Describing a trace as accounting for mic calibration is imprecise,
because it depends on your frame of reference. The key factor is
whether any trace represents an in-room response, or the response as
seen by your sound card ADC (after being filtered by an imperfect mic)
I think mortod's
mortod;230718 Wrote:
Hang on - both Inguz's test and my own observations make sense. DRC
applies mic calibration to the filter, not the TestConvolution which
means that all those frequency responses are without mic calibration
and *are* comparable.
I don't know, mortod. According to Hugh in
tonyptony;230390 Wrote:
I agree that's a head scratcher. If Hugh is right that means your
measured curve should look quite a bit different.
Hang on - both Inguz's test and my own observations make sense. DRC
applies mic calibration to the filter, not the TestConvolution which
means that all
I played with REQW and think I have my impulse response measured and
normal test convolution file in a form you can all get a good look at.
These measurements were generated using a flat target file.
As I mentioned earlier in the thread, Inguz DRC has improved my system
dramatically, but I'm
inguz;230255 Wrote:
Sounds like a great question :)
So I ran a little experiment. I set CrazyMicCal.txt to this (plus
some intermediate points):
100.00 -20
110.00 0
5010.00 0
5020.00 -20
and the result below (octave smoothed): green is the
Measured_Impulse_Response, purple is
inguz;230275 Wrote:
Well, there's no way to get a really true reading with an uncalibrated
microphone. And I don't have a calibrated mic. So that's my next
step...
I got mine calibrated a while ago and my advice would be: It should be
the first step! :P
--
tingtong5
I'm somewhat confused. Look at this graph again. Green is
MinimumTestConvolution and Red is MeasuredWithEQ. They are almost
exactly the same (apart from a 2db shift due to my global gain of -18).
Surely, if the TC line is with mic calibration and the measured without,
then they should be
htrd;230307 Wrote:
No, I think you were right all along.
drc believes that your mic is 20dB down at 5020Hz. drc sees that the
measured impulse response is relatively flat (green), and deduces that
your speaker+room system must be 20dB up. Therefore it produces a
filter with 20dB
That's great news, mortod. Yeah it will make REQW a bit more tweaky, and
users will have to be a bit more careful with it, but I think it will be
a big help to us. :-)
I'd actually vote for that feature to be able to be switched on or off
for each individual response which is imported. That way
tonyptony;228920 Wrote:
Hmm, okay, this may be a bit complicated, but if you look through this
thread (which I started, BTW! :-) ) you'll see some of us have posted
the frequency reponse curves of our own measurements and resulting
correction curves.
All of the WAV files created by Inguz
esbrewer;230657 Wrote:
If the only practical way for me to post is just buckle down and learn
REQW myself say so, and I'll try to find some time.
Esbrewer, I think you can use REWQ. Franckly speaking I struggled a
little bit to get my measurements and in particular to generate the
filters,
Chinanico;229961 Wrote:
- I didn't get my final license yet, so those were just trials
For consistency, don't - what you see in REW is also ignores the mic
calibration (BTW I am using a bk3spline target [you asked earlier], but
the lack of mic calibration makes that irrelevant when viewing REW
mortod;229964 Wrote:
For consistency, don't - what you see in REW is also ignores the mic
calibration (BTW I am using a bk3spline target [you asked earlier], but
the lack of mic calibration makes that irrelevant when viewing REW
responses as the RSSPL has such steep drop offs).
Are you
I might be mistaken, but I thought the ImpulseResponseMeasured and
various TestConvoultions were before mic correction (you only see the
mic correction data being used in prep.log, after the creation of the
ImpulseResponseMeasured). Plus my frequency responses seem to follow
the mic curve. The
mortod;230083 Wrote:
I might be mistaken, but I thought the ImpulseResponseMeasured and
various TestConvoultions were before mic correction (you only see the
mic correction data being used in prep.log, after the creation of the
ImpulseResponseMeasured). Plus my frequency responses seem to
Sounds like a great question :)
The Impulse_Response_Measured doesn't take account of the microphone
calibration file. It's just the measurement.
But the _TestConvolution files do. They're the result of processing
Impulse_Response_Measured with the DRC correction filter - they should
pretty
I know I should know the answer to this, but now what do we do?
--
tonyptony
tonyptony's Profile: http://forums.slimdevices.com/member.php?userid=3397
View this thread: http://forums.slimdevices.com/showthread.php?t=35615
Well, there's no way to get a really true reading with an uncalibrated
microphone. And I don't have a calibrated mic. So that's my next
step...
--
inguz
inguz's Profile:
tonyptony;230274 Wrote:
I know I should know the answer to this, but now what do we do?
I'm right there with you tonypony (at least I feel like I'm in good
company on this one), This has been in the back of mind since I got my
ecm8000 calibrated, but I was afraid to ask because I thought it
inguz;230255 Wrote:
the mic response is not 20dB down at 5020Hz, it's 20dB *up*).
No, I think you were right all along.
drc believes that your mic is 20dB down at 5020Hz. drc sees that the
measured impulse response is relatively flat (green), and deduces that
your speaker+room system must be
OK, I have attached three graphs:
1. Original measurement, plus Minimal TestConvolution,
2. Minimal TestConvolution plus Minimal measured,
3. Original measurement, plus 17 band low frequency equalisation
measured.
First of all, the Minimal RC predicted and measured are almost exactly
the same.
Chinanico, your curves look great, even accounting for the high
frequency reflections and that dip in the left channel. It looks like
you and mortod both checked the real response by doing a Sweep w/EQ and
regenerating the Impulse Response. mortod, are using using the flat
profile in the
I have never heard of TACT (for example) cautioning against RC with
certain speaker designs. The idea is counter intertuitive. The RC is
correcting what is arriving at the listening position. It really
shouldn't matter exactly how it got there...
I've heard TACT with Quad Electro's and Linn
Granted, the Isobariks were interesting (I remember hearing them at
Innovative Audio when they were in Brooklyn), but I guess it's not so
much that maybe an RC shouldn't be used in certain cases, but whether
there's enough information for omni (just to use an efficient term)
speakers in a
mortod;229692 Wrote:
I suspect that playing the sweep 'with EQ' ignores the equaliser
settings, and only allows testing of RC... could you confirm?
No, sweep with EQ runs the whole processing pipeline. I think you
may be hitting up against the inaccuracy of EQ bands at low frequencies
inguz;229908 Wrote:
No, sweep with EQ runs the whole processing pipeline.
Interesting. mortod, without having this info cloud your judgement,
what did you hear when you implemented the 11 band equalization?
--
tonyptony
I just made my first measurements with Audiolense, and wanted to share
my findings.
The 2 graphs below read as follow: first is the frequency response
curves (right= blue / left=red) made with audacity/inguz/REWQ, second
is frequency response made with Audiolense.
They don't really look
Chris, do you need to XO version to do things like creating target
curves or correcting small ranges of suckouts or bumps? Also, does it
do phase as well as gain?
--
tonyptony
tonyptony's Profile:
For tweeking the target curve and make notch filters in the curve you
don't need the XO version, even the basic version has it.
I must confirm with Brent at juicehifi but I think that to presonalize
the filter creation process like frequency limitation or windowing you
do need the XO version
Hi Chris,
As a matter of fact, I would appreciate if you could share some of your
experience as a tester, not sure how to import the Inguzprep measurement
files into Audiolense, or how to make measurements directly by sending
the signal into the squeezebox.
Thanks
Nicolas
--
Chinanico
Chinanico;229471 Wrote:
Hi Chris,
As a matter of fact, I would appreciate if you could share some of your
experience as a tester, not sure how to import the Inguzprep measurement
files into Audiolense, or how to make measurements directly by sending
the signal into the squeezebox.
I found in an older thread here a suggestion from Inguz which enables
some very crude 'notch' filters to be implemented.
http://forums.slimdevices.com/showthread.php?t=24519page=16highlight=Inguz+preset
I used REW to identify the peaks and amount of excessive gain, below
around 150Hz. I then
mortod, you wouldn't happen to have a before and after sweep that you
could post? Would love to see what happened.
--
tonyptony
tonyptony's Profile: http://forums.slimdevices.com/member.php?userid=3397
View this thread:
tonyptony;229515 Wrote:
mortod, you wouldn't happen to have a before and after sweep that you
could post? Would love to see what happened.
Planning to do one - just waiting for a suitable quiet (family out)
period to test.
--
mortod
I don't resist to post my new set of measurements. This time I centered
well the mike, and I didn't move it between measures. I went far into
the night to lower the background noise: Shanghai, where I live, is a
bustling city and that comes with its audiophile burdens...
The first graph (1/6
I think that the editing of the target curve and the limitation of the
filter frequency range can be easily done using the Audiolense software
from www.juicehifi.com . This software generates RC filters and has a
very nice user interface.
The target design is much easier than with DRC since the
Wow, AudioLense looks great. Seems to provide a whole lot of flexibility
in tweaking the correction curves. Might be worth it.
--
tonyptony
tonyptony's Profile: http://forums.slimdevices.com/member.php?userid=3397
View
tonyptony;229243 Wrote:
Wow, AudioLense looks great. Seems to provide a whole lot of flexibility
in tweaking the correction curves. Might be worth it.
That's right, might be the solution. But not free, on their web site
they say it has been tested with Inguz, anyone heard about it?
--
I would recommend anyone using Inguz/DRC to use REW to check just what
is going on. I had a lot of problems getting ImpulsePrep to produce a
good impulse response, due it turns out to a bug (now fixed) in
ImpulsePrep - but I could only tell a bad impulse response from a good
one by using REW. As
That 15dB suckout at 650kHz should be quite audible. Do you have access
to the Alan Parsons Soundcheck CD? The 1/3rd octave sweeps should show
that up if it's actually there.
I notice that there is a lot of HF roll-off, corrected and uncorrected
- is that right? You're way down at 10kHz... I
Phil Leigh wrote:
That 15dB suckout at 650kHz should be quite audible.
To a bat, maybe. :)
R.
___
audiophiles mailing list
audiophiles@lists.slimdevices.com
http://lists.slimdevices.com/lists/listinfo/audiophiles
Ooops - sorry I meant 650Hz
Thanks Robin :o)
--
Phil Leigh
You want to see the signal path BEFORE it gets onto a CD/vinyl...it
ain't what you'd call minimal...
...SB3+TACT+Altmann+MF DACXV3/Linn tri-amped Aktiv 5.1 system and some
very expensive cables ;o)
Phil Leigh;228966 Wrote:
I notice that there is a lot of HF roll-off, corrected and uncorrected -
is that right?
I assume that that is due to the graphs not being corrected for the mic
calibration? I'm using a RS SPL which rolls off steeply at low and high
frequencies. There's a feature in
Well, I have both REQW and TrueRTA, and now with Hugh having included an
uncorrelated Pink Noise test signal into Inguz I plan to use TrueRTA to
compare predicted to actuals. In my case I have a Behringer ECM8000, so
even if I don't use any cal corrections for the speaker there shouldn't
be too
Hmm, I shouldn't have said that. I just looked at the ECM cal file I use
when I use REQW to actually make measurements. It has some pretty
noticable falloff at either end of the band. The one used by DRC has
similar characteristics. I'll have to figure out what to do when I use
TrueRTA. The cal
Nice to see I've stirred up a hornets' nest. Muahaha. My work here is
done!
If a sound occurs in a room e.g. a note from a loudspeaker, the brain
is able to distinguish the directly radiated sound from the reflection
*if they arrive at sufficiently different times*.
This is one reason why
darrenyeats;228997 Wrote:
Omnipole and dipole advocates work on the above premise, namely
reflections happen so work with it not against it. I think a really
short Executive Summary would be by creating natural and familiar
acoustic conditions in the listening space, the brain can do its DSP
You may already know this, but an anechoic room sounds awful! It's
simply not natural.
Anyway this isn't all about reflected vs direct sound. There are other
issues such as standing waves and room modes.
I haven't tested RC with non-direct radiating speakers but I have
experimented with speaker
tonyptony;228991 Wrote:
I got a line on Audiocircle of someone who calibrates these mics, but I
never got a response to my email. I would like to get a cal for my own
mic.
I had my ECM8000 calibrated by Cross-Spectrum Labs (cost $50,
http://www.cross-spectrum.com/measurement/mike_meas.html).
I'm getting the feeling there's some slightly odd logic in play here.
When you listen to a good recording of (say) an acoustic guitar you are
hearing the sound of an instrument in a room. What you want to hear is
that instrument in that room, since that is what was recorded - okay so
it will
tonyptony;228789 Wrote:
Chinanico, I have seen very similar things in expected versus actual
response. In your case the extreme predicted curve shows a steady
downward slope as frequency increases, while the actual shows a similar
downward motion, but with changing slopes as you go from
Richard Vandersteen's take on EQ/RC :
Things are never as simple as they seem. We humans are very good at
differentiating between the source of sound and the effect of a room.
Put your best friend in 10 different rooms with very different sonic
characteristics and he or she will still always
Well Mr Vandersteen is entitled to his opinion - even if it is wrong!
:o)
His metaphor is wrong. The human brain+ear is specifically adapted to
speech recognition and also for discriminatary recognition of acoustic
spaces (the old bear in a cave thing)
Loudspeakers in a room are producing an
Chinanico;228785 Wrote:
IIs there a way to parameter things in order to have a very strong
correction in the bass (let's say below 200 Hz) and keep things
untouched above, and if there is, is it desirable?
I was going to raise the same question. This paper on RC by Meridian
suggests that one
I'm rather surprised at Meridian coming out with this... whats this
nonsense about trying to control the high frequencies will only affect
a tiny area?
That's BS. I don't need to be sitting anywhere special to hear those
HF's...I certianly don't need to be in a sweet spot.
I like Meridan. They
Phil Leigh wrote:
I'm rather surprised at Meridian coming out with this... whats this
nonsense about trying to control the high frequencies will only affect
a tiny area?
That's BS. I don't need to be sitting anywhere special to hear those
HF's...I certianly don't need to be in a sweet spot.
Robin,
Ah.
Well, HF's certainly are MORE directional...but I still don't
understand what directionality has to to do with the efficacy of RC?
If I'm off-axis it still sounds great to me. RC isn't better or worse
off-axis...it just is what it is (stereo imaging is a whole different
ballgame,
Robin Bowes;228865 Wrote:
haunyack wrote:
Richard Vandersteen's take on EQ/RC :
[snip]
I don't think he quite gets it.
Correct me if I'm wrong, but RC (hint: Room Correction) as I
understand
it doesn't really touch the original signal, it simply takes away the
effect of the room.
haunyack wrote:
Richard Vandersteen's take on EQ/RC :
[snip]
I don't think he quite gets it.
Correct me if I'm wrong, but RC (hint: Room Correction) as I understand
it doesn't really touch the original signal, it simply takes away the
effect of the room.
e.g.
Standard system:
I'm with mortod on focusing on notch filters rather than boost (maybe as
an option?). Inguz DRC has dramatically improved my system, but
clipping does become a problem. I am at -20db right now, and still
have significant clipping due to many newer recordings being mastered
too loud. My
esbrewer;228896 Wrote:
Everyone contributing to this thread has a greater command of technical
aspects of DRC than I do (I've probably failing to understand something
as I write this), but it does seem to me as if DRC further dulls the
already flat, compressed sound of newer recordings. For
I'd be happy to post the files, but I'm not quite sure exactly which
files you're referring to, and what form (if any) they need to be
converted to to be viewable. I really just followed Inguz's
instructions to the letter to set things up - but don't truly
understand what each component of the
Hmm, okay, this may be a bit complicated, but if you look through this
thread (which I started, BTW! :-) ) you'll see some of us have posted
the frequency reponse curves of our own measurements and resulting
correction curves.
All of the WAV files created by Inguz and DRC can be viewed in an
Hi Chinanico, glad you're liking it.
WAV is just fine. Actually it's the most CPU-efficient; if your music
is stored as FLAC then it goes through a decoding process first, before
processing, then gets re-compressed to FLAC afterward.
And -20dB gain to stop clipping is fine too.
--
inguz
Finally got my UL122, made my measurements, took a lot of time to
understand and run the DRC, and got a result.
The recording looks fine (well, maybe the recording level is a bit low
more around -12 than -6db), but it looks like strange results as far as
the response (first is filtered, second
Doesn't look strange to me. My uncorrected response looks similar. It
looks like you're using REQW. Did you convert the impulse response to
Stereo 16 bit PCM to get it to load into REQW? If so, is the measured
response the average of the two channels?
You should probably also post it with 1/6
tonyptony;228164 Wrote:
It's really not as bad as it looks. Most of the curves you see in
magazines are at 1/6 or even 1/3 octave smoothing. I've always wanted
to ask Hugh which level of smoothing best matches the interpretation of
what we hear.
There's no single easy answer (unfortunately)
I just wanted to say, of course there is more than the curves... the
difference in the sound is very big. I think on most tracks it is to
the better, much more clarity and lisibility. On pink noise, the
difference is impressive, but the only problem is I don't have much
more experience of
Octave success... Phase Response at 5ms. Guess this is why things sound
so much better with InguzEQ. My listening room really is an acoustic
nightmare.
What does the big spike at 1.25Khz mean?
+---+
|Filename:
muski;227721 Wrote:
Octave success... Phase Response at 5ms. Guess this is why things sound
so much better with InguzEQ. My listening room really is an acoustic
nightmare.
What does the big spike at 1.25Khz mean?
Sound only travels around 1.5m in 5 ms, which means (if I understand
Sorry, my description of the graph was wrong. It is Unsmoothed phase
response, using a 5ms Blackman window.
--
muski
SB3-Bryston BP25DA-Bryston 4B SST-Wilson Watt Puppy 7
Transporter via XLR-Headroom Max Balanced Headphone Amp-Balanced
AKG701s HD650s
IR1 impulse center shouldn't be 1. Can you email me the
LUncorrected.pcm and the Impulse_Response_Measured.wav files? Thx -
--
inguz
inguz's Profile: http://forums.slimdevices.com/member.php?userid=1139
View this
P.S. Line 81 of createdrcplots is:
# Creates the oversampled impulse responses
ir1os = rf * iupsample(ir1([ic1-irp:ic1+ira]),rf,rfl,rcf,rat);
--
muski
SB3-Bryston BP25DA-Bryston 4B SST-Wilson Watt Puppy 7
Transporter via XLR-Headroom Max Balanced Headphone Amp-Balanced
AKG701s
Oops, typo = wrong parameter...
octave:12 createdrcplots(ru,01, L Uncorrected,rc,-1,L
Corrected,./tmp,R);
second parameter should be -1, not 01...
Sorry,
muski
--
muski
SB3-Bryston BP25DA-Bryston 4B SST-Wilson Watt Puppy 7
Transporter via XLR-Headroom Max Balanced Headphone Amp-Balanced
Hugh, I've been doing quite a bit of measuring, using a number of
different tools. Recently I've been frustrated with the apparent lack
of consistency between TrueRTA and REQW, which I use to look at the
average of the two impulse response channels which I get out of the
measured impulse response
Were these measured with the microphone in the exact same position?
Even with a very small offset from the original measurement position,
you'll notice comb-filtering effects at high frequencies. You can hear
this really clearly if you move your head a few inches while playing
mono pink noise -
inguz;227475 Wrote:
Were these measured with the microphone in the exact same position?
Mic was in the exact same spot the whole time.
inguz;227475 Wrote:
To measure the in-room overall frequency response with both channels
(rather than a single speaker), you're best to use uncorrelated
Latest ImpulsePrep also adds some subsonic filtering. So if you re-run
against your sweep measurements, you should be able to see better
long-reverb graphs.
--
inguz
inguz's Profile:
So as I noted either in this thread or another, I've been recapturing
the sweep responses. I looked at the impulse response and compared it
to the one I got in May, which was the one I used to create the filters
I'm using now. I don't understand why there is a difference like this
between the
1 - 100 of 181 matches
Mail list logo