• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Low- and Mid-Frequency Test Signals for Evaluation of Room Modes (and other things)

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,021
Likes
4,408
Location
Berlin, Germany
Hi,

Recently, I promised I'll present my trusted test signals for evaluation of room modes, finding best positions for speakers and listener, check for distortion/rattling etc.

This was a project I did 20 years ago before EQ'ing was commonplace, let alone full digital "room correction".

An important feature was that I tried to highlight the temporal distortion that comes with room modes, something which cannot easily be evaluated with steady-state sine test signals. These are specifically designed burst sequences, ranging from 27.5Hz to 440Hz, spaced in quarter-note distances.

The usage was optimized for CD-players (preferably capable to display CD-text info) with remote control because other digital playback sources also weren't commonplace.
Using a remote control of any sorts which can be blindly operated, to quickly navigate through the tracks with instant playback starting after a "next" or "previous" select, is paramount for convenience and success. You don't want to waste any effort on navigation since the main task is well-focussed concentrated listening.

Download Link (containing a bunch of FLAC files, some instructions and a GIF for a CD index sheet).
Source and executable files (Windows) for the C programs used to make the original .WAV files are available on request.

Information :

LF_TEST V1.0
-- Low Frequency Sine-Burst Test Signals for Room and Loudspeaker Evaluation --

HOW TO USE THE CD (or the signals in general)
=============================================

MOST OF ALL :
BE CAREFUL WITH THE PLAYBACK VOLUME!!
YOU MIGHT EASILY DESTROY YOUR WOOFERS.
ALL SIGNALS ARE RECORDED AT FULL LEVEL!!


1) GENERAL:
A CD-player with CD-Text option for track name and a remote control is most convenient for usage of these test signals. The CD uses all 99 tracks allowed by the CD standard which any CD-player should be able to handle.

You should be in a relaxed state of mind and the environment should be quiet but not dead silent -- some background random noise is helpful to better detect the structure of the modulations because of the psychoacoustic compression effect; the noise gets masked away by the LF signal in the specific patterns thus making them more readily audible.

To make yourself accustomed with the CD's signals, you might want to listen to some tracks at first with headphones to memorize the target sound of the patterns more easily. Start with a moderately low frequency like track #23 (49Hz).
Make sure your speakers are really off before turning up the volume of the phones!! Don't overdo it as you might damage your phones also and your ears will thank it, too.

For the first time of usage you may want to reserve at least an hour for testing.
If you happen to get bit by the "burst bug" you may find yourself listening for many hours, especially when you start to move your speakers and listening position around.


2) Now start with the first track, the noise, to set a rather low listening level and for general verification that your setup is working well. You might already notice some frequencies "sticking out".

3) Play track #2, the sweep. The track has 3 seconds of silence at the beginning, afterwards the low rumbling of 27.5Hz starts without warning.
DON'T TURN UP THE VOLUME because you think at first it were too low to hear the low tone. Wait and watch woofer's motion before you decide to turn it up!

The sweep is timed such that the current position (in seconds as displayed) by the player indicates the track number of the corresponding sequence for the current frequency we're at, in the sweep. So if you hear a drone or a suck-out or any other interesting stuff at, say, 00:23 and 1:12, the corresponding detail sequence tracks are track #23 and track #72 (60+12).

The sweep is in mono, so you will likely find the phantom image to heavily go off-center at specific frequencies, an indicator of some L/R asymmetry in your setup (typically not easy to nail down what it causes it, though).


4) THE STRUCTURE OF THE BURST SEQUENCES, for each specific frequency:

There are 97 sequences on the CD, ranging from 27.5Hz to 440Hz in quartertone steps, 24 sequences per octave. On the CD index sheet the frequencies their musical notes are given.

Each sequence is based on a burst of several cycles with raised cosine fade-ins/fade-outs of 2 periods. This allows a gradual start and stop of the burst, to avoid detracting clicks.

Several parameters are varied systematically within each sequence:

  • the total burst length starting with 1 second then getting shorter, but no less than 6 periods (2 fade-in, 2 normal, 2 fade-out).
  • the gap in between (in multiples of periods) plus the phase relationship between the bursts (so that the next burst starts either in phase with a room hangover or in counter-phase by making the gap half a period longer).
  • the interchannel polarity. Either mono or one channel "out of phase".

The sequence consist of four blocks:
  • each block starts with long bursts reaching steady state, and long gaps in between the burst, then gap length is successivly shortened.
  • the next block basically indentical but has counter-phase gap lengths (x.5 periods instead of integer multiples).
  • finally, these two blocks are then repeated again but with one channel inverted, L and R signals are out-or-phase.

If you have a .WAV editor load the first burst sequence (27.5Hz) to get a visual idea what is happening, additional to the proposed headphone listening check.


These variations create a big pool of possible signal combinations to check the effect of room modes. The inverted parts are for double checking mostly if you have modes with strong sideways components.

This can also be checked with one speaker disconnected which you should try as well.

NOTE: If you have mono subwoofer(s) there will be NO SIGNAL if the frequencies are low, for those out-of-phase section (the second half of the four block set per sequence).

Feel free to walk along the room and put your ear at specific positions (including different heights).
After some time of usage (say after two or three sessions) you will be familiar with the LF response pattern in both level and temporal (time-domain) aspects. The burst sequences will develop many interesting patterns, even some kind of specific rhythm.

Compare with headphones at times to re-check how it really should sound like. You will then note that some music with specific bass rhythms may sound corrupted, out of time and without any articulation, while others may sound extremly tight and punchy. But it may not be the musicians that are responsible for that perceived effect!

Also you will know the musical notes where your setup has problems and it will become clear why songs sound lame in the chorus because the bass line hits a drop-out region.

Finally you will start to move your speakers around as well as the listening position including height of both to see if you can make the overall response more benign and even. Quite often positions can be found where this can be achieved and even 10cm (4") can make a big difference (for symmetry problems especially, as the geometric center line of room does not need to conincide with the acoustic center when L/R walls are different in stiffness.

Take care that you check and verify other aspects of reproduction like reflections, stereo imaging, tonality in the midrange etc, especially if you have a two-speaker-only setup (no subs). You may and up with better bass response but otherwise degraded performance from where you started.

Don't get too disappointed if you don't seem to find any good bass setup and smack-center mono image, with real music it will not sound that bad, typically.

Also if you decide to use EQ to flatten the percieved response don't overdo it. EQ'ing is a big topic of it's own that would bloat this text even more. Same goes for the theory about modes, the internet forums are full of information about this and books are also available.

Have fun!
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
554
Likes
515
Little Xmas present from Klaus :) Thanks a lot!

Just as an addition, similar bursts can also be produced with REW's "Generator":

Screenshot 2021-12-24 at 12.20.44.png
 
OP
KSTR

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,021
Likes
4,408
Location
Berlin, Germany
@markus,
Indeed, with REW one can create bursts and burst sequences that are somewhat similar to what I'm using here.
Yet, the key point of my sequences is the well-defined spacing between the bursts giving either an in-phase or an out-of-phase continuation, plus being dynamically varied in this spacing, plus having them on 97 frequencies. With REW and a wave-editor you'd need days of manual generation and editing to get anywhere near of this.

In-phase Continuation detail, with the shortest possible spacing (0 cycles):
1640346752680.png



Out-of-phase Continuation detail, with the shortest possible spacing (0.5 cycles)::
1640346833833.png
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
554
Likes
515
With REW and a wave-editor you'd need days of manual generation and editing to get anywhere near of this.
I wouldn't suggest anyone to do this. That's why we have your test signals. REW's signal generator can only be complimentary if one wants to play around and listen for the audible effects of room modes.

So thanks again for posting these, especially the .wav's. Should make it very easy for anyone to experience the effects of room modes.
 

restorer-john

Master Contributor
Joined
Mar 1, 2018
Messages
8,729
Likes
24,233
Location
Gold Coast, Queensland, Australia
Great post.

Is there an easy way to 'select all' the files to download at once? (I don't do dropbox other than via my browser)
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,043
Likes
1,422
Location
Canada
Great post.

Is there an easy way to 'select all' the files to download at once? (I don't do dropbox other than via my browser)

No need for a dropbox account. When you click on Download in the upper corner of the screen:

1641571049710.png


You should get the direct download link in a compressed zip file.
 
OP
KSTR

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,021
Likes
4,408
Location
Berlin, Germany
Exactly,
Just ignore/close the popup windows, click "download" and wait for the download manager of your OS to start as seen in above screenshot.
File size at ASR is limited to 4MB so I would have to split up the data into fours segment to host it here directly. Can do that if people find this more convenient...
 
OP
KSTR

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,021
Likes
4,408
Location
Berlin, Germany
I did, but I don't think my feedback is usefull cause in my treated room and a single subwoofer those signals don't seam to reveal anything
You mean the shortening in-phase time gaps sound as clear and pronounced as the anti-phase gaps, no smearing of the in-phase gaps at any frequency?
And no center image shift along the sweep, and out-of-phase channel burst signals do perfectly cancel?
Summed up: perfectly free of any room modes and perfectly symmetrical in everthing? Basically identical in timbre and stability to headphone playback?

I very strongly doubt this, because, IME (and that's 20years) this only achievable in very anechoic space (outdoors, large halls).
No normal room ever could come close, including well-designed recording studios -- even with DRC applied.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
2,453
Likes
1,333
Location
Salvador-Bahia-Brasil
You mean the shortening in-phase time gaps sound as clear and pronounced as the anti-phase gaps, no smearing of the in-phase gaps at any frequency?
And no center image shift along the sweep, and out-of-phase channel burst signals do perfectly cancel?
Summed up: perfectly free of any room modes and perfectly symmetrical in everthing? Basically identical in timbre and stability to headphone playback?

I very strongly doubt this, because, IME (and that's 20years) this only achievable in very anechoic space (outdoors, large halls).
No normal room ever could come close, including well-designed recording studios -- even with DRC applied.

I replied from memory. I will try again this weekend
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
554
Likes
515
Anybody have used the test signal yet?
I have and would anybody encourage to use them. They are highly effective for getting a basic idea how a room and speakers perform by using just your ears.
 

tw99

Senior Member
Joined
Mar 13, 2019
Messages
377
Likes
802
Location
West Berkshire, UK
I've just tried to download the files but am having some trouble. I can download the .ZIP from dropbox, but then when I try to extract any of the .FLAC files, Windows explorer gives me an %x80004005 unexpected error. I've tried this on a couple of machines and extracting to different places.

I've also tried the cygwin unzip utility and that says for example " skipping: 392.0Hz.flac need PK compat. v6.3 (can do v4.6)"

Any pointers to a tool that can actually unzip this ?
 

tw99

Senior Member
Joined
Mar 13, 2019
Messages
377
Likes
802
Location
West Berkshire, UK
I've just tried to download the files but am having some trouble. I can download the .ZIP from dropbox, but then when I try to extract any of the .FLAC files, Windows explorer gives me an %x80004005 unexpected error. I've tried this on a couple of machines and extracting to different places.

I've also tried the cygwin unzip utility and that says for example " skipping: 392.0Hz.flac need PK compat. v6.3 (can do v4.6)"

Any pointers to a tool that can actually unzip this ?

@KSTR

Answering my own question, 7Zip worked to extract them.

The files are very useful and thanks for posting. I'd previously been using a Youtube sweep posted by someone else on here, but your files have much more resolution. Having a listen to them tells me that my SHD Dirac setup is working reasonably well compared to the sound with Dirac disabled. Mid-40s and mid-60s Hz are still a bit uneven, but I'm not sure that I can do much better given room constraints.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,043
Likes
1,422
Location
Canada
I have listened to all tests signals in full with and without sub (with and without EQ) as well as through my ATH-M50 headphones. So far, this is preferable to REW's signal genrator. I already know that I have some balance drift issue between my (couch MLP) left and right speakers when playing in full-range unequalized. However, I want to re-listen to the tracks later at a much quieter time so I'll be writing more about my experience later.

Quick summary: playing the sub as it is by default in full-range, there doesn't seem to be any drifting apart from hearing expected room-induced amplitude peaks and dips. Though, as I mentioned, with the LR speakers played unequalized (and to a much smaller extent equalized), there is some center/channel drift...
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,043
Likes
1,422
Location
Canada
Okay, so here's a little more detail of my experience. Note that my room is acoustically treated and the placement of speakers is optimized.

First with the ATH-M50 headphones (review link of the M50x)... bass is very clear and centered throughout 27-440 Hz. Doesn't seem completely "flat", but the overall response is fairly even. I must have been listening too loud as I seem able to hear the diaphragm mechanically push and pull in the lowest frequencies -- later my ears were ringing and tinnitus was exacerbated.

Inverted portion of the test signals sounded louder and coming from right side.

Next with equalized bass-managed front speakers...

Below is a not so completely accurate (but close enough) simplified diagram layout:

1644850024824.png

The actual concrete foundation behind the drywall extends to the right, wherein the total area of the basement roughly equals more than twice the depicted room size. The dividing walls and ceiling are "flexible" and insulated which helps absorb a significant amount of the energy.

Bass is very clear and mostly, strongly centered. It also sounds "smoother" and maybe "less articulate" than the one coming from the sealed headphones. There remains some slight bit of channel drift but not much. Dips can definitely be heard, but they are fairly narrow and sound does not fall off like a bottomless cliff -- meaning that for the most part the sound remains decently even. Around 300 Hz the amplitude rises a bit louder than the rest of the response. No ringing can be heard.

As for the equalized subwoofer by itself... it sounds mostly flat overall except for a few dips that are noticeable above 100Hz -- but again, nothing excessive.

However, with the un-equalized subwoofer, there is one mode in particular that stands out above the rest:

1644850509833.png


BTW, at the start, with the sub channel playing in solo, the inverted phase portion (of the summed stereo signal) at moderate level (50% software player volume) is completely inaudible. Past 50 Hz, one can just barely hear it positioned at center.

If one is not careful in lowering the volume before playing the test signals, the amplitude and the reverberation experienced in the room is overwhelmingly powerful and disturbing. The actual 20-30Hz ringing in the room is still short in perspective, though, with a calculated modal RT60M that varies somewhat, fluctuating to as low as 0.65 to as high as 1 second. I haven't determined what the cause of the fluctuation is -- the sweep settings, noise levels, and doors/windows being opened or closed appears to be all partially involved.

SUB ONLY
1644851631316.png


ALL CHANNELS (L+R sweep in 4.1ch DSP upmixing mode):
1644851719414.png

The expected RT given the size of the room should be around 0.14 s -- with my room treatment in place, it's much closer to 0.1 s

The glaring spike around 20-30 Hz is rather not bad. I've heard from an acoustician that many rooms can measure closer to 2-3 seconds.

Using sensible spectrogram settings, we get a more intuitive and easy to understand picture of how "problematic" the room's subwoofer bass response is:
1644852477109.png

Truth be told, EQ will not get rid of the ringing effect of this particular mode esp. near the rear wall couch MLP -- but it can help reduce it significantly enough to where it becomes a "problem" of no import.


The unequalized transfer function of the main front speakers are disparately affected by the room and positioning,

(FDW 15 cycles)
1644853891527.png


Channel drift is easily noticeable using the swept test signal and peaking most annoyingly around 300hz. But this only starts to begin maybe around 127 Hz (but it moves/varies along the way) using the individual frequency centered signals. To be frank, it is still not what you would call "super obvious" in practicality with most media playback content unless you really listen hard for it: male vocals and some bass instruments sometimes feel biased to the right. But in general, under 250Hz or so, the summed bass response dominantly sounds centered accompanied by variation in amplitude level.


With the help of a sub and some EQ (no boost greater than 3dB -- to my left+sub channel), I hear a more balanced full-range bass response.

1644855351072.png 1644855358325.png

What little channel image biasing/drift that remains at the MLP is negligible.


Full sweep LF test signal sounds centered and relatively flat apart from a few small dips in amplitude with bass management and equalization.

The inverted phase half of the tracks is more apparent with the mains playing along, and as mains SPL level rise -- gets louder. Early on that part is not always so obviously centered, but as it gets louder, it becomes more solidly anchored to the right side.


Also if you decide to use EQ to flatten the percieved response don't overdo it. EQ'ing is a big topic of it's own that would bloat this text even more.


To keep it short, I've learned to keep my own EQ "mild" in order to maintain a more even power response in the room. But here's an example:

(FDW 15)
1644856031650.png


That's the response I get with the same EQ used for the couch MLP (closer to rear wall) when listening alternatively at the desk in front (closer to middle front of room and significantly off-axis). Obviously, mids and highs are attenuated a lot with some notching, but the sound is still very good -- actually, even better for background listening where concentration and focus must be placed elsewhere.


1644856176457.png



My listening setup(s) is very modifiable (2.0, 2.1, 5.1, 7.1) and we can actually create a more headphone-like experience if the monitors are rotated so they are perpendicular to the wall and directly firing to each ear:


1644856765370.png


The room and subsequent EQ applied, while good, is still not perfect. BUT my listening tests and measurements indicate that it's indeed (perhaps unusually) very well-behaved.



More pictures, just because...
1644858464106.jpeg 1644858486649.jpeg 1644858498408.jpeg 1644858582977.jpeg 1644858535166.jpeg 1644858541593.jpeg 1644858550861.jpeg

Added internal air-spacing in between in those DIY bass panels like a sandwich to make them behave maybe a bit more like membrane traps -- to some degree. Behind and below the burgundy panel are also 4 individual 6-inch rockwool containing "limp-bags" just using wrapped and (hopefully) "air-sealed" garbage bags. However, most the frequency response is the result of the room itself, and the DIY traps and 4-inch thick acoustic foam wedges just improves the decay behavior making it more even/flat and slowly rising to the bass as seen in the RT graph.
 
Last edited:
Top Bottom