What measuring techniques do you recommend to ensure that my monitors are calibrated correctly for my room?

There are a variety of measuring techniques that are available today but some are more suited for measurements in recording studios and other relatively small room environments than others. Summarized below are most of the commonly available techniques and some tips to ensure that you can make a reliable measurement with the tools that you have available:

  • 1/3rd Octave real time analyser (RTA)

    Pink noise is played through the monitor and a, typically 1/3 octave band, LED or graphical output waveform is displayed. The sound field in the room should become stable before conclusions are drawn about the measurement. The results obtained with this method will be same as that from an MLS system with 1/3 octave smoothing and an MLS signal length of over 300 ms. This is because the modal excitement is the same in both cases. However, this means that the reflections and resonances in the room become more significant in the measurement than they really are with real music signals as there is no direct-to-reflected energy selectivity. This technique is also susceptible to noise. This is a quick and relatively reliable measurement technique if a good quality measurement microphone is used and is especially suitable for bass to midrange balancing.

  • Spot frequency using sine wave and a sound level meter

    Use a simple sine wave signal generator to excite the monitor at specific frequencies and then measure the level using a sound level meter. This technique is not only susceptible to noise but it is also inaccurate and time consuming. Again this method measures the steady state conditions of the room which emphasizes the room effects. This method is only suitable for low frequency modal evaluation in highly damped rooms.

  • Warble tone spot frequency

    This technique tries to reduce the room effect by modulating the test frequency slightly. It is less susceptible to noise than the RTA and sine waves techniques. It still suffers, to some degree, from the effects noted above but gives consistent results.

  • Swept sine wave

    The traditional method uses a sine wave that is swept from low to high frequency together with a graphical output showing the frequency response. The speed of the sine wave sweep smoothes the frequency response. This method still suffers from low noise immunity and an over emphasis of the room effect but does give consistent results. However, sweeps in conjunction with FFT analysis are very powerful and allow to exclude the room effect and possible noise by windowing, just as is possible with MLS. Moreover, in contrast to MLS, they also allow to completely exclude harmonic distortion.

  • Time Delay Spectrometry Analysers (TDS)

    A fast sine sweep is used as the signal. A swept tracking filter is used before the analysis stage so that interference from noise is minimized. The room effect is also greatly reduced as, by the time a reflection reaches the microphone, the tracking filter has moved to a different test frequency. The width of the filter and speed of the sweep can be used to include or exclude as much of the room effect as required. This is analogous to time windowing an impulse response. This method suffers from a loss of accuracy at low frequencies (due to the tight 'time windowing') and the measured response is automatically smoothed depending on the sweep speed and filter width. This gives user selectable trade-off between the room influence and frequency resolution. Provided the parameters have been set properly this technique can be used to measure monitors in studios both reliably and consistently.

  • Two-channel FFT (Fast Fourier Transform) Analyser
    Using an impulse (a very short and loud sound, e.g. a gunshot or bursting balloon)

    This method is commonly used in measuring the room acoustics of concert halls. The transfer function (frequency and phase response) of the monitor in the room is calculated from the Fourier Transformation of the monitor's impulse response. The impulse response can be time widowed to remove the effect of reflections but low frequencies become inaccurate if too short a window is used. Unfortunately, it is difficult to produce a perfect impulse as all monitors have a limited maximum sound pressure level due to the cone excursion limits and amplifier power rails. Also, it is difficult to measure an impulse as it contains very little energy.

    Using pink noise (or even music!)

    A signal with energy at all frequencies (this already presents a problem as you have to be sure this is the case) can be used as an input signal and the transfer function calculated from the cross-correlation of the input and output. This suffers from accuracy errors, as the input signal is, by its very nature, random. To overcome this, the input signal should be played for a long time and time averaging is used to reduce the noise element of the signal. This generates a new problem; the long term output of a drive unit is limited by voice coil heating so the signal can only be played at a given sound pressure level for a certain length of time (the relationship is approximately exponential - loud sound, short time). To overcome this problem the level should be reduced but this reduces the signal-to-noise ratio of the monitor in the room. The other major problem is that this is a steady state measurement and so includes all the room reflections in their entirety.

  • Maximum Length Sequence Analysers (MLS)

    This technique uses a self-generated periodic, pseudo-random noise signal, which is cross-correlated with the measured signal to give the impulse response, and hence transfer function, of the monitor. The signal is called a Maximum Length Sequence and sounds a bit like, and has similar energy content properties to, white noise. This signal has been optimized for FFT algorithms due to its 1-bit on/off nature. This signal contains equal amounts of energy at all frequencies so the signal can present problems to monitors at high frequencies (high energy content and low crest factor). Adding a pre-emphasis filter to the signal and de-emphasis filter to the resulting measurement can overcome this problem and give an excellent signal-to-noise ratio in the room. As the signal is a pseudo random signal, i.e. not totally random and therefore unknown, the effects of signal noise are greatly reduced. Time windowing can be used to limit the effect of room reflections and averaging can also be used to further increase the signal-to-noise ratio. This measurement technique is reliable and repeatable.