Audio Measurement Tips - Listen, Inc.

Electroacoustic measurements rely on many different audio test stimuli. This brief summary of the various options guides you towards the best stimulus for your measurements.

Stepped Sine Sweep
A stepped sine sweep plays a sine wave, one frequency step at a time with a user-defined frequency range and resolution. SoundCheck’s stepped sine sweep, the ‘Stweep’ is optimized with an integer number of cycles at each frequency step and continuous phase between steps for a smooth transition from frequency to frequency. This minimizes transient ringing in the device under test for faster and more repeatable measurements. Stepped sine waves can be configured with different resolutions and averaging times at high and low frequencies to optimize the trade-off between stimulus duration, background noise and therefore, test time and accuracy. The stepped sine sweep is one of the most widely used audio test stimuli for basic electroacoustic testing because it offers good noise immunity and provides acoustic (e.g. frequency response, phase, THD), perceptual distortion, and electrical (e.g. impedance) measurements with a single sweep.

Stepped Sine Sweep (Stweep) Audio Test Stimulus

Log Sine Sweep (AKA Frequency Log Sweep, Log Chirp or Farina Sweep)
A continuous log sine sweep from low to high covers every frequency in the chosen range. The same time and energy in each octave achieves a good signal-to-noise ratio at all frequencies. The sweep rate is slow at low frequencies for accurate measurement, but increases with frequency for a fast overall test speed. SoundCheck was the first test system to implement simultaneous simulated free field measurements of harmonics using the Log chirp stimulus, just months after Angelo Farina’s landmark AES paper on the subject in 2000.

The log sine sweep is a good alternative to a stepped sine sweep for production line loudspeaker measurements, and is also common for fast time-selective measurements of fundamental response and distortion. Used with a Time Selective Response algorithm it provides simulated free-field measurements for free-field analysis of a loudspeaker without an anechoic chamber.

Log Sine Sweep (AKA Frequency Log Sweep, Log Chirp or Farina Sweep)Audio Test Stimulus

Log Amplitude Sweep
A Log amplitude sweep is similar to a stepped sine sweep except a single frequency is chosen and swept logarithmically through various levels.

It is useful for evaluating how a device performs at different stimulus levels (linearity tests) and for testing devices with compressors, limiters or automatic gain control (AGC). It can also be used with advanced analysis algorithms to evaluate characteristics like THD vs. stimulus level.

Log Amplitude Sweep Audio Test Stimulus

Noise
Both pink and white noise can be used for audio testing. Pink noise has a continuous frequency spectrum and equal power per constant percentage bandwidth, e.g. equal power in any one-third octave band. It is used with constant percentage bandwidth filters such as the RTA. White noise has a continuous frequency spectrum and with equal power per unit bandwidth, for example, equal power in any band of 100-Hz width. It is generally used with FFT analysis.

Noise test stimuli are often used where sine waves cannot be used, such as devices with on-board signal processing e.g. codecs.

Pink Noise Audio Test Stimulus

White Noise Audio Test Stimulus

Multitone
A multitone stimulus consists of multiple sine tones at discrete frequencies played simultaneously with equal amplitudes and random phase. All the frequencies are present at the same time, in contrast to the stepped sine sweep where the frequencies are present sequentially. All frequencies are covered in a very short stimulus, providing very fast measurements. Multitone is good for measuring frequency response, but the number of simultaneous frequencies in the stimulus makes it impractical for measuring other characteristics such as harmonic distortion, limiting its use.

Multitone test signals have applications in simple, rapid production line tests, for example measuring the frequency response of microphones.

Two Tone Stimulus

A two tone stimulus plays two simultaneous tones. In a non-linear system, these generate intermodulation distortion, output frequencies that are not part of the stimulus and have no harmonic relationship to the original signal. There are two commonly-used types of two-tone stimulus. An intermodulation (IM) two-tone stimulus is a stepped sine sweep superimposed on top of a fixed frequency tone. The fixed tone is usually at a low frequency, and the sweeping frequency is generally five times greater than the fixed tone frequency. A difference frequency (DF) two tone stimulus is composed of two simultaneous sweeping tones separated by a specified frequency interval which may be a fixed difference or a fixed ratio.

Measuring distortion using a two-tone stimulus presents an alternative to harmonic distortion measurement in which the distortion products are kept close to the stimulus frequency and typically fall outside of the masking curve. This makes artifacts more clearly audible and avoids the under/over estimation that can sometimes occur with higher order harmonic distortion methods. IM distortion is generally used for broadband devices such as full range drivers and microphones. DF distortion is used for measuring band-limited devices such as hearing aids.

Two Tone Intermodulation Audio Test Stimulus

Two Tone Difference Frequency Audio Test Stimulus

Speech and Music Based Signals
Speech and music based stimuli enable devices to be tested with real-world test signals. This is particularly important for audio devices with built-in non-linear signal processing. In SoundCheck, any WAV file can be used as the stimulus, making it simple to use music, real speech, or artificial speech. Additional software features such as active speech level control enable these signals to be used for a variety of different purposes including communications testing to industry standards.

Applications include phones, codecs, hearing aid compressors and more. Additionally, these test signals are valuable for creating calibrated background noise for testing smart devices, headphones and wearables.

Speech (wav file) Audio Test Stimulus

Composite Stimulus Signal
Composite stimuli allow a conditioning signal to warm up the device or open up an activity detector before playing the stimulus signal. They are essential for testing smart devices, where it can be necessary to create a stimulus that consists of a wake word plus a command, or to prepend a stimulus with a trigger tone to enable precise alignment of stimulus and response. SoundCheck offers total flexibility to create customized composite stimuli, and also permits the user to precisely select which sections of the stimulus should be analyzed.

Compound Audio Test Stimulus consisting of voice activation followed by a test tone

All these stimuli are available in SoundCheck. Learn more about SoundCheck’s features and functionality.

Audio measurement troubleshooting can be challenging. All too often we plug and unplug things, tweak settings and re-test, often changing several variables at once. A consistent and logical approach greatly accelerates the process. Here, we share the process that our experienced team of support engineers has developed to assist customers. Follow these 10 audio measurement troubleshooting tips to quickly get to the root of your problems.

Check your cables. Cables are the single biggest cause of setup problems! Make sure all your cables are properly connected and in the right place. Next substitute alternate cables one at a time – cables often get broken inside or damaged, even though they pass visual inspection. Make sure you understand the different types of cables and when you should use each – particularly where they look similar, e.g. single-ended and balanced cables. It’s a good idea to label your cables to make it quick and easy for anyone to replicate your measurement setup and quickly fix things if anything becomes unplugged.
Check everything is plugged in! It sounds obvious, but if there is no output signal or not what you expect (e.g. it just looks like noise), check that everything is plugged in and turned on – not just the computer, but all audio interfaces, amplifiers, hardware, microphone power supplies etc.
Run a self-test. Always run a self-test before making measurements to confirm that your measurement system and cables are working correctly. This facility is usually built into your measurement system and confirms that all components are working as expected.
Make one change at a time and document everything. Don’t move cables, tweak the sequence and change the microphone position all at once – you won’t know what the cause of the problem was! Change one thing at a time, and keep a checklist of what you have done (this will be useful if you need to call customer support).
Check your audio interface configuration. If your audio interface is incorrectly configured, nothing else will work. Ensure sound monitoring is switched off to prevent feedback, and make sure you are using the full dynamic range. Try making a loopback (output to input) measurement on the audio interface using the appropriate SoundCheck self-test. It should have a flat frequency response. If your frequency response looks noisy it is likely that you have insufficient gain. You should have unity (0dB) gain when looped back on itself and very low THD. Latency should be consistent and repeatable and must be ≥ 0 Record Delay. There are many places where audio interface settings can be incorrectly configured – check your hardware setup, ASIO control panel and mixer (if applicable), device front panel (if applicable) and Windows audio device settings.
Calibrate your signal paths. Calibrating your signal paths is critical to accurate measurements and you should always have a multimeter and an acoustic calibrator handy! Calibration should be carried out at regular intervals as well as when you first run the test, particularly if there is a change in atmospheric conditions. Good measurement microphones and electronics are typically very stable but loudspeakers e.g. mouth simulators are generally very non-linear and their performance changes with temperature. It’s always a good idea to warm up a loudspeaker before testing it by playing pink noise for example at a reasonable level. Factories should re-calibrate every day or at the beginning of every shift.
Measure twice. Always measure something at least twice on the first measurement to make sure that you get the same result. This helps identify problems with the setup or background noise. It’s far better to discover this on the first measurement than at the end of a day of data collection! A golden unit, e.g. a demo speaker that you have measured many times, is a good sanity-check if measurements appear unexpected. A golden unit is advantageous over calibration as it also takes into account the fixturing and test environment.
Check your units and your resolution. Make sure you are measuring in your intended units. For example the difference between dB Pa and db SPL is 94dB. If you are not careful you could blow something up. It’s also important to check your resolution. Lower resolutions provide faster measurements, but it’s important to ensure that you are using enough. If you keep increasing the resolution until the curve doesn’t change, you can easily identify the lowest acceptable resolution to test at.
When in doubt, look at your recorded time waveform. Use your software’s oscilloscope function to see if your recorded waveform looks noisy, has drop outs, or gets cut-off too soon. Look at the peaks of the recorded waveform and Max FSD in the memory list to see if it looks unexpectedly flat – it may be overloaded and clipping. These problems are very difficult to see in the frequency domain.
Minimize background noise. Background noise is the biggest cause of unrepeatable measurements, so minimizing this improves your test environment. Easy ways to minimize background noise impact include positioning the microphone closer to the source, increasing the test level, increasing the duration of the test signal, and using repeated averages of the test signal. Doubling the averages or signal duration should increase the signal to noise ratio by 3 dB.

And a bonus #11, submitted by Chris Struck at CJS Labs.

11. Check that you have sufficient record padding. The length of the record padding must be longer than the acoustical propagation delay from source to receiver, plus the maximum group delay, plus any digital signal processing delay, plus the length of any conditioning signal — to ensure that the entire stimulus signal is captured. If only a part of the stimulus signal is captured, a result will be recorded, but this will not correspond to the correct response.

We hope you found these audio measurement troubleshooting tips useful. Follow our blog for more handy hints for audio measurement. And if you have something you think we should add, please let us know!

Download the poster for your audio lab wall:

Audio Test Stimuli: 8 Audio Test Signals and When to Use Them

Audio Measurement Troubleshooting: 10 Time-Saving Tips

The Value of Big Data

Audio Basics: Phase Response

USB Headset Measurement

Overcoming Challenges of Measuring Distortion Audibility in Automotive Audio Systems

Audio Basics: Frequency Response

Practical Automotive Measurements Seminar

Celebrating 30 years: Listening – The Secret Of Our Success

Automotive Test Sequence Including BSR, Max SPL and Frequency Response

Listen Introduces AudioConnect 2, a New Portable Dual-Channel Audio Interface

SoundCheck 21 Brings Enhanced Loose Particle Detection, Multichannel FFT, Metadata, Sequence Protection and more

ABOUT LISTEN INC

Registration required

ABOUT LISTEN INC