SpeakerMeasure User Guide
Automated AES75 Measurement | Step-By-Step Instructions
How to use SpeakerMeasure to Test Loudspeaker Maximum Linear Sound Levels
IN THIS GUIDE
Estimated reading time: 13 minutes
SpeakerMeasure is a software tool that automates the AES75 test procedure for measuring loudspeaker loudness.
It was developed by Michael Smithers, who served on the AES SC-04-03 Working Group on Loudspeaker Modeling and Measurement that developed the AES75 standard.
SpeakerMeasure with AES75 is available both for macOS and Microsoft Windows & is available free of charge to anyone interested in making consistent AES75 measurements. Download SpeakerMeasure here.
Get the AES75-2023 standard
Before You Start: Things to Know
Music Noise
The AES75 test procedure uses a unique test signal called Music Noise. You will need to download the Music-noise test signal from the AES website in order to use SpeakerMeasure.
Tech Notes
SpeakerMeasure uses continuous transfer function (TF) and coherence measurements. Common controls (like FFT size and overlap percentage) are provided, and the TF is not conditioned by the coherence. The FFT’s use a regular Hann (or Hanning) function window (and not the frequency-varying time windows found in some commercial, live TF analysers).
Although TF phase is not displayed, compensation delay is calculated and applied automatically, whenever audio playback is started.
WARNING & DISCLAIMER
This software facilitates driving loudspeakers at very high sound levels to test their limits. Hearing protection should be worn at all times while testing.
The software has undergone significant testing but is provided ‘as-is’. Eclipse Audio is not responsible for any damage to equipment or hearing that may result from use of the software and/or possible bugs in the software.
Users of this software should already be familiar with the AES75 measurement procedure. A copy can be obtained from the Audio Engineering Society.
Equipment Setup
The following diagrams show example setups for passive and active loudspeakers. A high-quality audio interface is required. (On Microsoft Windows, the audio interface must have ASIO drivers.)
SpeakerMeasure Setup
SpeakerMeasure can be used in either light or dark mode. Audio IO settings and metering are configured in the column on the left. The test procedure and settings fill the remainder of the screen.
SpeakerMeasure interface in Dark Mode
Audio IO
In Audio Device, select a device driver and set the buffer size to 256 or 512 samples.
In Outputs, select the soundcard output channel that is feeding the amplifier or the active loudspeaker.
At least one microphone is needed for the TF calculation. Optionally, SpeakerMeasure can use a second microphone for SPL measurement (instead of using a dedicated, physical SPL meter).
In Microphone Inputs, select the soundcard input channels that are connected to the microphones. (If using only one microphone, the other mic input can be ignored.)
The TF reference input is locked to Stimulus. This means the TF calculation will use the internal stimulus signal which, for the AES75 test, is the loaded WAV file; presumably Music Noise. More details below.
Microphones
In the Microphone box, select which microphone will be used for TF and which for sound level or SL. Both TF and SL can use the same microphone and therefore be set to the same mic number.
For the SL microphone, enter the distance (in metres) of the microphone from the loudspeaker. Note: metres = feet x 0.305.
Microphone Calibration
The Calibration tab holds the relationship between dB SPL and digital level.
If the values are already known for the audio interface and microphones being used, then check Edit under each column to enable the text fields. Enter the values, then uncheck the Edit checkboxes.
If values are unknown, check SPL Calibration and follow the instructions shown.
Stimulus
The AES75-2022 test procedure uses a unique stimulus signal called Music Noise (download it here).
In the Stimulus box, load the 48 kHz Music Noise WAV file. (Only 48kHz is supported.) The file statistics are calculated and displayed.
Optional HP and LP filters can be enabled and are applied to the stimulus during the test. When enabled, the file statistics shown include the application of the filters.
FFT Settings
At the heart of the TF calculation are Fast Fourier Transforms (FFTs).
The AES75 procedure recommends averaging eight successive TF measurements with a 50% time overlap between each measurement.
The total averaging time window is calculated from the number of averages, the overlap percentage, and the FFT size.
How to choose the FFT size?
The frequency range that will be monitored during the testing is set using the Min freq (Hz) and Max freq (Hz) settings in the Level Increase Stop Conditions box.
Min freq (Hz) is also used to calculate the minimum analysis time interval (or FFT length) suitable to achieve at least 1/6 octave analysis resolution in the TF.
Here, a min freq of 100 Hz requires a FFT size of at least 0.087 seconds. Comparing this to the 0.683 s in the previous picture, 0.683 is clearly greater than 0.087, so the FFT size of 32768 samples (at 48 kHz) is ok.
Note the min and max frequencies are displayed on the plot as vertical orange lines.
Test Procedure
The AES75 test procedure consists of two parts: the Linear Response Check and the Level Increase Test.
PART 1: The Linear Response Check
Here the loudspeaker is subjected to a modest voltage level at which the loudspeaker is assumed to be performing normally and with low distortion.
STEP 1
With Linear Response Check selected and the Output gain (dB) set to approximately -40.0 dB, press Start and adjust the amplifier gain (for a passive loudspeaker) or the input gain (for an active loudspeaker) such that the loudspeaker is approximately 20 dB below its expected rated RMS power or continuous SPL.
STEP 2
Make a note of the Vrms from the voltmeter.
STEP 3
Next to Provisional Linear Response, press Capture. This captures the TF at the present level and holds it on the plot as a light blue line. (Shown below.)
Note: A warning will be displayed if the coherence squared is <= 97%, between the min and max test freq.
STEP 4
Increase Output Gain (dB) by at least 3 dB.
STEP 5
Make a note of the Vrms from the voltmeter.
STEP 6
Next to Confirmation Linear Response, press Capture. This captures the TF at the present level and holds it as a green line. Two other lines appear; a thin red line denoting 2 dB lower than the confirmation response and a thicker red line 3 dB lower than the confirmation response.
- The 2 dB lower line represents the threshold for the contiguous 2-octave TF droop stop condition.
- The 3 dB lower line is the threshold for the TF point stop condition.
Note: Warnings will be displayed if the coherence squared is <= 97% and/or the “Confirmation Linear Response” deviates from the “Provisional Linear Response” by 1 dB or more.
STEP 7
Press Stop.
PART 2: The Level Increase Test
Here the loudspeaker is subjected to a slowly rising voltage, with a growth rate set by the Level growth (dB/min) setting.
When the Provisional Linear Response and the Confirmation Linear Response have both been captured, the Level Increase Test Start button becomes enabled.
To begin, press Start.
The output level will slowly begin to rise at the rate set in the Level growth (dB/min). The Current output gain (dB) displays the instantaneous output gain. Unweighted A and C weight SPL measurements are also shown, both for the mic SL mic position and referred to a distance of 1 metre.
As the output level is rising, the TF is continuously being checked (according to the AES75 procedure) for:
a) any 2-octave continuous frequency region of the TF level that might be >= 2 dB below the Confirmation Linear Response,
b) any point of the TF level that is >= 3 dB below the Confirmation Linear Response, and
c) any point of the coherence2 that is at or below 91%.
If any of these conditions occur, the output gain is reduced by the Fail backoff level (dB) amount, the SPL infinite measurements are reset and a timer is started to hold the output level for the duration set in Max Sound Level Duration. The timer’s elapsed time is shown next to Elapsed @ max.
Whilst the timer is running, make a note of the Vrms from the voltmeter.
At the end of the Max Sound Level Duration time, the test stops. The plot window and the displayed SPL can be copied to the clipboard or saved to an image file using the lower-right buttons.
Progressive Level Backoff
While the timer is running, the TF is still being checked for the a), b) and c) conditions above. If any of these occur, the output gain is again reduced by the Fail backoff level (dB) amount and the timer is restarted.
The goal is to hold the loudspeaker at a constant level over the Max Sound Level Duration and with the loudspeaker performing within the limits specified in a), b) and c).
Plot Zoom & TF Normalizing
Two plot settings provide a closer view of the TF
Zoom freq. to passband expands the plot to show only the frequency range from Min freq (Hz): to Max freq. (Hz): in the Level Increase Stop Conditions settings box.
Normalise normalises the plot by the Confirmation Linear Response so that the two red threshold lines are flat and so that the live TF shows the amount that the TF has changed, relative to the Confirmation Linear Response.
Configurations
The application loads previous settings when opening, and saves all settings at close. However, specific combinations of settings can be saved and loaded using the Configuration box.
Preferences
Number of Audio Buffers
If there are noticeable glitches in the sound playback, increase this number.
Generally, 50ms of buffering should be more than enough for stable performance. (With a buffer size of 256 samples and a sample rate of 48 kHz, 50ms corresponds to ~9 buffers.)
Enable advanced AES75 settings
When checked, previously greyed out settings in the FFT and Level Increase Stop Conditions boxes become available. The defaults are per the AES75-2023 standards document.)
SpeakerMeasure FAQs
No; the tool has been designed solely for automating the AES75 test procedure for determining maximum linear sound levels. Its not a general-purpose acoustic measurement tool (unlike Rational Acoustics’ Smaart software).
You’ll find this option within ‘Preferences’. After changing the setting, restart SpeakerMeasure.
Eclipse Audio haven’t created a video tutorial but the screenshots and information in the User Guide should give you all the information you need to use SpeakerMeasure. This tutorial from SydAudCon is also helpful.
