LARSA is a powerful impulse
response measurement and analysis tool. It is ideally suited for measuring
loudspeakers in rooms, multi-speaker setups, car audio systems, home theater
systems or larger, installed loudspeaker systems. It is an easy way to profile
and equalize your loudspeakers in your listening space. Quickly make an impulse
response measurement in your listening space, generate frequency curves and
energy decay curves. Then optionally generate filters to help match the room
response to the target curve and generate a file ready to import into many
popular DSP platforms. (Coming Soon as an in-app purchase)
LARSA helps to:
* Measure and profile the
frequency response and energy time curve of your loudspeaker system at numerous
positions in your room.
* Quantify the presence of room
mode interactions in your playback system.
* Match subwoofer systems to
full-range mains speakers.
* (Optional, coming soon)
Automatically generate filter settings ready to be imported into compatible
hardware devices.
* Quantify magnitude response
differences between speakers or difference channels of a multichannel speaker
system.
* Aids with sound field
management, EQ and bass blending.
LARSA does not:
* Correct for large anomalies in
magnitude response due to improperly set up systems, or deficiencies in
playback hardware (i.e. it will not allow you to get 20 Hz bass extension from
a small monitor loudspeaker.).
* Correct
loudspeaker rattles, extended or inappropriate room reverberation or
audible echoes.
* Correct for improper
loudspeaker placement, wiring or signal flow.
The Basics
Setup
Basic Measurement
Spatial Average Measurement
Measuring Multiple Loudspeakers in the
Same Location
IR Windowing Options
ETC Graph Options
Automatic Filter Design
Home
Theater Loudspeaker Delay Measurement
LARSA is an acoustic measurement
tool that mathematically compares a reference log sine sweep signal to that
same signal measured through a transducer and room. This is useful, in that by
subtracting the measured signal from the reference signal, what remains is just
the impact of the loudspeakers, other devices in the signal chain and any room
interactions.
There are a
number of ways to record this signal. With our iAudioInterface2, you
will plug in the line output of the interface into the input of the loudspeaker
you wish to test. You will plug in your measurement microphone into the
microphone input of the interface and set up the microphone in the location
where you wish to measure the response.
With our iTestMic, the
configuration will be the same, except you will use the line output of your iOS
device to feed the measurement signal into your system.
Finally, you can use the internal
microphone to capture the measurement signal, although for best results, a
measurement microphone should be used.
The first step you need to
perform is to take a reference measurement to determine the latency of your
system. Tap the wrench icon and in the ETC Graph setup screen you will see a
latency box. Set this to 0.0s to start.
Tap done and place the microphone
as close as you can to the transducer you wish to measure. Set the gain of your
microphone preamp by tapping the gear icon, and then microphone setup. For
complete calibration and gain adjustment instructions, please refer to the
instructions in the settings screen.
Tap the measure button to take
your first measurement. A log sine sweep signal will be recorded and through a
process known as deconvolution, the resulting impulse will be generated.
To set the latency, switch from
the analysis curves screen to the ETC screen by tapping the second grey icon on
the bottom middle of the module. The ETC, or energy time curve, shows how
energy is decaying in your room. First, assure that the 1.00s option is
selected in the ETC graph. Use a single finger drag to slide the cursor to the
peak of the impulse response and observe the time readout. This is the
reference latency of your system. Tap the wrench icon again,
and type the time value you observe into the latency field.
Now take another measurement in
the same reference location. This time, tap the 0.020s button on the ETC curves
graph. Your new measurement’s peak should be 0 ms. If
it is not, adjust your latency settings until your close mic is measuring 0 ms delay and 0 ft distance. Once that is set, as you move
around your room and make other measurement, the delays and distance
differences will be meaningful.
There are two basic ways to use
this module. The first is to make individual measurements of different
loudspeakers or measurement locations around your room. The other is to make
multiple spatially-averaged measurements of the same
loudspeaker.
We will start with the first case
where you wish to measure, for example, a small studio monitor in 1) a close
microphone configuration (as close to the transducer as possible), 2) at the
listening location and then 3-8) at multiple locations in the diffuse field.
You have two options for sweep lengths, which can be selected in the setup
screen. The 0.5s sweep gives you good resolution and a quick measurement time.
The 1.5s sweep gives you a bit more resolution in the lower frequencies, which
is useful if you are measuring large loudspeakers that have an extended low
frequency response or subwoofers.
The sweep measures from 10 Hz -
22 kHz, so please be careful of the signal level sent to your loudspeakers.
Protect your hearing and your equipment. If you are measuring for an extended period of time, please use ear plugs. In the audio section
of the setup pane, there is a test tone option, which will toggle a sine tone
for level setting (this signal is also available on the main screen). Assure
that the signal is intense enough, but not too loud that it will distort your
microphone input. (about 75-85 dBA at 1 m is a good starting point)
In addition to the sweep length
setting, you can also select if you are using a balanced out or a mono out and
select the input source.
Back in the main screen, we are
ready to take a measurement. Tap the 1 test number and position the microphone
in the close microphone configuration (as close as you can get to the
transducer). Now tap measure. In the analysis curves graph, you will see the
frequency response of your measurement. The currently selected measurement in
the test number will be displayed as a blue trace. Test numbers that have a
stored IR will be displayed in green text, and open locations will be displayed
in red text. The AVG shows the average frequency response of all measurements.
In the setup screen there is an option to turn on show all plots. This will
toggle being able to view all measurements you have made in a single graph
(with the currently selected measurement shown in blue) and looking at just the
current measurement.
The frequency response graph can
be smoothed. In the setup screen there are options for the frequency graph.
Curve smoothing options from none to full-octave smoothing are available. You
can also manual set the ranges for the dB (Y) axis, or
frequency (X) axis.
Next, move you microphone to the
listening location (usually located in the sweet spot between your stereo
loudspeakers). Repeat the measurement process. Notice how the intensity of the
frequency response has decreased because you are further away from the
loudspeakers? In LARSA, none of the responses are normalized, so you can see
the level differences and time difference in multiple measurements around your
room.
Move over to the ETC curves,
assure that 0.020s graph time option is selected and notice how you can see
that the second IR that has been measured is further back in time than the
close mic IR. This is because there is a delay introduced by the distance from
the loudspeaker to the microphone. Drag your finger onto the plot the measure
this delay. Line up the vertical cursor with the second peak and you will see
the ms delay and the delay expressed in distance.
Move your microphone to the third
measurement location and repeat the process. You can repeat this process for up
to 8 measurement locations in the room. To store your results, tap the folder
icon. For complete save / recall instructions, please refer to the info in this
screen.
If you have inadvertently made a
spurious measurement, or wish to clear the memory in
the test number boxes, highlight a test, and tap the garbage can icon. The
module will prompt you the clear this test (the one selected), all tests, or
cancel. All tests will remove all of your measurements
from memory (1-8), no matter which one is highlighted.
A useful technique for
characterizing how a loudspeaker sounds in a room is to measure a spatial
average around the sweet spot.
The technique here is to measure
IRs around the listening location and take an average of all 8. This helps to
smooth out modal behavior and gives you a reasonable impression of the
frequency response that you are actually hearing from
your loudspeaker. This technique is useful for setting tone controls on a
pre-amp or receiver, or general equalization of your system. This is also a
useful technique for setting EQ setting present on many active studio monitors
or DSP equalizers.
Start again by setting your
latency as instructed above. Tap test number 1 and place the microphone where
the close to the listening position, at ear level with the microphone pointing
up. Continue the same process for the remaining 7 measurements adjusting your
measurement in a random fashion around your listening location. If you listen
in a chair, keep the measurements within a 1 foot
radius in all dimensions around your listening position. If you listen on a
couch or similar, the radius can be 3 feet or so. After you are done, tap the
AVG button. This should be a reasonable approximation of the frequency response
of the loudspeaker in your room. Note, you are not looking for a flat response
here. Most systems that sound perceptually "good" have a slight
elevation of the response from 20 Hz - 200 Hz, are flat to 1 kHz and then have
a gradually dropping high frequency response towards 20 kHz. To see just the
average measurement, and not the individual ones, enter the setup screen and
turn off the show all curve switch. Then tap done and tap the AVG Test Number.
Another measurement technique
that you can use LARSA for is to make measurements of multiple loudspeakers at
the same location. We will take the case of measuring a 5.1 surround sound
system in the sweet spot of a home theater.
Again, you will start by making
the reference latency measurement. Then place the microphone in the sweet spot
with the diaphragm pointing up, positioned at ear level. Tap test number 1, and route the signal to the left loudspeaker. Measure the
response. Tap test number 2, route the signal to the center channel and measure
the response. Continue measuring the right, right surround, left surround and
subwoofer in the same way.
Now you can compare the frequency
response of each loudspeaker as well as the arrival time by inspecting the ETC
curves screen.
After an IR is recorded, it is
windowed in the time domain to reduce errors in the frequency response due to
the abrupt stopping and starting of the IR sound file. Windowing simply
smoothly removes some of the beginning and end of the IR. Since most loudspeaker
systems are being measured in rooms with relatively short reverberation times,
this has minimal impact on the resulting ETC analyses. There are three options
for time windowing. They can be adjusted in the setup screen. The options for
adjusting the time of the window are available for short, medium
and large.
You also have the option to
disable the time windowing, which can be useful for inspecting the noise floor
and signal to noise ratio in your room.
In the setup menu there are
options for adjusting the display of the ETC graph. You can manually set the dB
scale (y-axis) and time scale (x-axis). Here, you can set your system latency,
as discussed in the basic setup section. Options are here is well to express
the x-axis in units of time or distance. Finally you
can turn on ETC Smoothing. This applies an exponential moving average smoothing
function to the time response. You can turn this off, or
have a time window of 10-40 for the smoothing.
Filter design is available as an
in-app upgrade in the LARSA module. It is accessed through the setup menu.
There is a switch to turn on the generation of equalization filters. The idea
of the filter designer is to take the measured frequency response of your
loudspeaker and design inverse filters to flatten the response at the listening
location by attenuating room modes. There are a number of
commercially available digital and analog devices, which can perform this
filtering.
To start, tap the Target EQ Curve
text and load in a house curve. This is the curve that you want to match your
measured response to. The target EQ curve uses the same protocol as the
reference curves available in AudioTools. Full details about generating and
loading in reference curves are found in the reference curve info section.
Tapping use default curve will load a standard reference curve.
LARSA supports automatic filter
generation for the miniDSP 2x4, generating a text
file of filter coefficients that can be loaded directly into miniDSP's desktop plugins. If you have a 2x4, tap the miniDSP 2x4 text. You can also choose a generic 4 or 5-band
filter generation option, which will give you a text file of the frequency,
bandwidth (Q) and gain for each of the filters. These can be input manually
into a digital or analog equalizer.
You have options to specify a
maximum boost and cut in dB of the designed filters. Generally, it is a good
idea to avoid large boosts. This is a good way to introduce distortion, reduce
headroom and damage your loudspeakers. Good starting points might be +3 dB and
-10 dB. Also increase the volume of your system carefully after adding
equalization filters.
When perform filter design is
turned on, addition traces are shown in the analysis curves graph. Tapping the
text will turn on or off the traces. The before trace is the raw frequency
response of the system. The target is the reference curve that the filters are
trying to turn the raw response into. Predicted shows what the response should
be after being processed by the equalization filters. Finally
the filters trace plot the magnitude of each of the peak filters being used in
the equalization. Tapping the magnifying glass icon will bring up a table,
which shows the gain, Q and frequency of each of the
filters.
LARSA Loudspeaker Delay Test (LDT) provides a solution for
obtaining the real-world delay time of the loudspeakers in a surround system,
compared to the reference loudspeaker. To use it, you will need the LARSA
module and the Apple TV Surround Generator app, with either the 5.1 signals or
the Atmos test signal package installed. LDT is a free feature, once you have
the LARSA module installed.
LDT requires the special LARSA dual-sine chirp signals that
are in the Apple TV Surround Generator app.
LDT works by analyzing the dual-swept sine signal, which
sends one chirp to the reference (left) loudspeaker, and the other to the
loudspeaker chosen for testing, offset by a fixed amount. LARSA uses the offset
time to pick out which signal comes from which loudspeaker, and run them both
through the same analysis, and determine precisely the
difference in the initial delay time.
This process ends up being able to determine the loudspeaker
delay difference, including all electronic processing delays, DSP delays, or
any other sources of delay, since we are using the actual acoustic waveforms.
This process is very repeatable, noise immune, and can resolve differences with
a resolution of 0.5ms.
Setting up and Calibrating Loudspeaker Delay Mode
Turn on Delay Mode on the LARSA settings page, and select
External Chirp, and 0.5s Chirp.
We recommend setting the Left channel delay in your receiver
to the largest distance that you expect to see, based on the distance to each
of the other loudspeakers.
This will usually give you the best results, as then the other channels will be
delayed relative to the Left channel.
Make sure that the sub will not be engaged while you are
testing the main speakers. If you cannot turn off bass
management, or are unsure, just unplug any subs during the main speaker tests.
Start the Surround Generator app on Apple TV 4k and select
ATMOS or 5.1 and the Delay signal. Pick a the Left
loudspeaker so that we can check the delay calibration.
Tap Measure, then start the test signal on Apple TV. It may
need to loop to engage the processor. Both chirps will be output from the Left
channel.
Now you should see 2 plots on the main screen. Switch to the
ETC screen (tap the > button on the screen to switch plots), and you
should see the delay time, which should be 0ms, or very close to it. The two
plots on the screen are showing the two deconvolutions which have been
computed. If the time is not very close to 0, open the settings page and scroll
down to the +/- control that shows the Delay Trim Samples. Adjust this value to
increase or decrease the calibration time and go back to the ETC plot and check
the delay time. Repeat this until you get a value close to 0ms. In many cases
leaving the trim value to 0.0ms will be the correct setting.
Measuring Loudspeaker Delay
For these tests, make sure the test microphone is centered
in the room, at the main listening position. Remember that moving the
microphone 1 foot right or left can make a 2ms change in delay. You want to
optimize the delay for the main listening position for the most immersive
experience.
Now change to another loudspeaker, for example Center, on
the Apple TV Surround Generator. Tap Measure on the LARSA app and click the
selected speaker to start the signal. Note that if you are using this function
in the HAA app, you can link to the Surround Generator and control it directly
from the HAA app.
You will see two plots, and a new delay number which is the
delay to the loudspeaker under test. It may be quicker to write this number down
and collect all the delay times before changing the delays in the receiver or
processor.
Repeat the test for the rest of the loudspeakers and write
down the results.
When you get to the subwoofer, follow the same procedure.
The sub should be able to reproduce enough of the sine chirp to get good
results, although there are other methods that may be used for sub time
alignment.
Now you can adjust all of the delay
times in the receiver by adding or subtracting the numbers and go back and run
the test again and observe the results. If needed, go back and again adjust the
receiver delay times.
Tips for Best Results
Use a fairly loud test signal,
since we need to be able to clearly find both of the sine chirp signals.
If you have multiple subwoofers that have independent delay
time controls, disconnect all but one and test them separately.