Audio Measurement

Sedona, Arizona, USA (Photo by Yuichi)

Measurement of a speaker system and / or room acoustics is very important. Some one says that it is a kind of try and error activities. It may be true. However, when we ask the doctor in case we feel something wrong with our body, first the doctor will gather possible physical data such as blood pressure, body temps, blood analysis then check specific area by X-ray or CT or.... Finally finds out problem and its root cause, and then takes some actions to recover.

Same as above, we have to know the physical analysis data for speakers, a system and a listening room first. This is very key to get our goal effectively. There are many tools available for audio measurement as of today. Many freeware or shareware can be found in the internet.


CLIO by Audiomatica Audio Tools by Studio Six Digital
Platform : Windows Personal Computer Platform : iPad/iPhone/iPod touch
Here, I can explain about audio measurement data examples referring Audiomatica's CLIO system which I have been using. The system is very well designed and provides comprehensive functions. There is no need to have all of CLIO functions for all audiophiles, but you can refer those for each needs. Some of the measurement screen shots are displayed here.
(http://www.audiomatica.com/home.htm
The mobile equipment is easy to move the measurement environment anywhere and anytime. But it used to have limited measurement functionality in the past.
The Audio Tools by Studio Six Digital realized very powerful measurement facilities on the iOS equipment. One of the questions is how to get qualified input signal from the Microphone. Coming soon!

(http://www.studiosixdigital.com/)
See below for CLIO Click here to AudioTools

Frequency Response
Frequency Response Measurement (Swept sine wave method)
---------------------------------------------------------
There are several methods to measure the gain-frequency response of the speaker system. Traditional way is to sweep sine wave signal and give the signal to the speaker system through a power amplifier. Then, captures sound pressure output by a microphone. The trace is the SPL-frequency response curve shown in the left picture. The graph is often called simply as a "frequency response".
The frequency response property in audio is something like a blood pressure of human body. Therefore, this is a fundamental property of the audio system. Sinusoidal sweep can also finds out 2nd, 3rd, 4th, 5th, total harmonic distortion (THD) and phase property on the basic curve. Left graph shows basic frequency response (Black) and some of harmonic distortions (Red+Green). The data shown is for a single corn speaker system property.
Frequency Response
Frequency Response Measurement (MLS Impulse response method)
---------------------------------------------------------
One of the other way to measure the frequency response is to use MLS impulse. MLS impulse is generated by a periodic pseudo-random binary sequence. This can also be applied to the speaker system. Then, captured by a microphone. Analysis is made by using original sequence. Then, a impulse response with higher S/N ratio can be obtained. This time domain data can be converted to frequency domain data by FFT, which is a frequency response data shown in left graph which is for a 3way system.

In reality, MLS method is very useful. Because we have many analysis capabilities available in the post process once MLS impulse response is obtained as a part of pre-process.
Frequency Response
Frequency Response Measurement (RTA method)
--------------------------------------------------------
RTA(Real Time Analyzer) will give us real time frequency response measurement capability. The pink noise is applied to a speaker system. A microphone gain is now FFTed in real time with octave band such as 1/6 or 1/12 or so. Advantage of this method is that the measurement is done in real time. Therefore, we can change environment such as listening (mic.) position change, speaker location change during measurement. The graph changes real time. So, it is very useful to find a best position of listening or alignment of speakers for instance.

The white noise can also be used for the frequency measurement. In this case FFT analysis (not octave band) can be used.
Impedance Measurement
Impedance Measurement
-------------------------------------------------------
Impedance of a speaker system also includes a lot of speakers information. To measure the Impedance, a power resister (such as 1 ohm/1W) is inserted in between a power amp and a speaker voice coil. Swept sine power is applied to the voice coil. Induced voltage on the resister is admittance of the voice coil. Inverse of admittance generates impedance.
If we drive the speaker system by higher output impedance amplifier such as a tube amp, speaker (load) impedance is very key to determine sound coloration. SPL is enhanced in the area of high load impedance area as shown in the Experiment page in this Web.

Phase property is also measured, and MLS method can also provides these measurements.
Thiel Small Parameters
Thiel Small Parameters (T/S) Calculation
-------------------------------------------------------
Thiel Small parameters are needed to design a enclosure of low frequency speaker driver. To generate T/S value, we have to measure impedance with two different environments. One is in the free air condition. And the other is with additional load on a diaphragm (Corn in most cases). There are two ways to apply corn load. One way is to put physical weight on the corn. The other is to apply air load on the corn which is to mount the driver in a sealed box.
Then, we can generate two impedance curves. Black curve (left) is in the free air and red one (right) is with corn loaded. We can choose one of the convenient way depending upon the diaphragm material or construction of the driver. Picture below shows air loaded way.
On top of above measurement, some static value have to be input to the CLIO system. Which includes DC resistance of the voice coil.

Here, Thiel Small parameters are calculated and displayed. Fs, Qts and Vas are the most important parameters especially for a bass reflex enclosure design.
Inductance at 1KHz and at 10KHz is deferent. This will give us information about the voice coil inductance compensation parameters in higher frequency zone.
Directivity Measurement
Directivity Property Measurement
---------------------------------------------------------
Directivity property is also important to find out good listening area. It also gives us information to decide a sound absorption materials on the wall or floor or ceiling.
Below is a 2D hand made turn table fixture. MLS impulse response is taken by every 5 angle from -180 to +180.
Left curves give us a classical directivity property of loudspeaker by specified frequency. This graph shows 2.5K, 5K, 10K, 16K and 20KHz properties
.
Here is different visualization method of a directivity properly. Horizontal axis is frequency and vertical is gain. From front to rear shows the deviation from the center axis by degree. There are variations of setting for different visualization. However, this is to equalize center axis gain to 0dB. Therefore, we can purely analyze directivity property free from original SPL. This example is s super tweeter data which has excellent performance.
This another way of visualization of a directivity property. Higher SPL is displayed by red colored and lower one is by blue. Horizontal orientation is frequency axis and vertical axis displays from +180 (bottom) to -180 (top).

This is good to know the sound distribution in the space in one chart. The classical polar pattern is sometimes misleading because only selected frequency directivity curve(s) is(are) shown. This color mapping method gives all seamless information.
Impulse Response
MLS Impulse Response
-----------------------------------------------------


MLS is generated by known random sequence as described before. This is why we can generate impulse response with high S/N ratio.
Impulse itself has lots of audio information in it. First response is direct sound from a speaker. 2nd, 3rd and 4th arrivals are reflected sound from walls/floor/ceiling. Thereafter, lots of reflected sound arrive. We call this as a echo or a reverberation.

These arrival can be observed on the impulse wave, if we see it carefully. Wave expanding capability is useful for the observation.

Cumulative Spectrum Analysis
------------------------------------------------------

This is one of the analysis to visualize frequency response as time dependent manner. Cumulative Spectrum Analysis is done by series of FFT calculations of the impulse response. Each FFT is shifted by certain time. So, we can observe frequency response by shifting time to right.

This specific data shows a vent output (SPL) of a vented low frequency enclosure. This shows how the output gets increased and decreased after impulse has applied. Please note that the horizontal axis is for frequency, the vertical axis is for gain (SPL) and from rear to front is the time axis.
Acoustic Parameters
Acoustic Parameters (RT60)
------------------------------------------------------

When MLS impulse is obtained in a listening room, captured impulse contains room echo. Room echo is very important factor for sound coloration. Reasonable echo is needed but less or many echo generates not natural sound. Echo is explained by reverberation time (RT-60) value and can be calculated the impulse data.

It is often said that 0.3 to 0.6 sec will give us adequate RT depending upon music type (classical or Jazz or other music) and individuals.

Left data shows the energy decay (red) and strait line (black) indicates integration of the energy. RT60 (sec) is calculated time spent from 0dB to -60dB. In reality, -60dB is very small to measure. Therefore, RT60 is by RT20 x 3 or RT30 x 2. In the left spread sheet, all RT60 values are shown and graphs by every frequency are displayed.
Wavelet Analysis
Scalogram Visualization
---------------------------------------------------
Cumulative spectrum decay (by FFT) gives us a good visualization idea of sound. However, problem is that it is less resolution of time domain. There is STFFT (Short time Fast Furrier Transform) to resolve above issue. But STFFT creates another problem which has less resolution in frequency domain because of smaller number of FFT points.

Wavelet analysis compromises above two approaches. However, analysis algorithm is totally different from FFT. In this wavelet analysis, it scans wave data with certain frequency filter with window function. Then, it finds out frequency contents in each time line in the waveform.
However, this method is good for higher frequency area. But in lower frequency area, the resolution looks worth because the frequency axis is logarithmic expression in general. It depending upon resolution setting, anyway.
Wavelet visualization is displayed by scalogram or 3D manner. The horizontal axis is time expression and the vertical axis is for frequency expression in scalogram. Red color is high SPL and Blue is low.

Left picture shows a time alignment of multiple drivers system. Above shows that the sound from the tweeter arrives faster than the woofer sound. This is why higher than 5KHz sound is about 1msec faster than lower 5KHz sound. After inserting a digital delay line in tweeter side, alignment of both drivers matches each other. Now we can obtain linear phase speaker system in multi-driver way.
When the source MLS impulse response is obtained in a listening room, important reverberation information is also included in the impulse. In this case, we can analyze how long and what frequency factors the reverberation contain. So, we can have better idea to improve room acoustics and can create effective actions to be taken by using the scalogram display.
Linearity
Linearity of speaker output
-------------------------------------------------------


Linearity of a driver output is measured. Harmonic distortion measurement gives us some information about input situation status on the driver. However, in this test we can find out saturation point very clearly. We can apply specified frequency signal to the under test driver. We have to be careful about radiated sound energy sometimes impacts human ears. In low frequency zone, driver diaphragm generates large displacement. Therefore, we also have to notified about the diaphragm damage. To avoid these problems, we can gradually increase the input signal.

Audiomatica can be searched by "CLIO" or "Audiomatica" CLIO provides many other measurement functions other than above.
Top page