At Golden Ears, the measurements of electrical - as opposed to acoustic - signals from cell phones, amplifiers, DACs, and audio players are conducted using a program called RMAA (RightMark Audio Analyzer). RMAA is a PC-based program which uses the PC’s sound card (specifically the Analog / Digital Converter) to analyze test signals.

RMAA (Rightmark Audio Analyzer)

RMAA is capable of analyzing many different characteristics, including the frequency response, noise level, dynamic range, total harmonic distortion (THD), intermodulation distortion (IMD) + noise, and stereo crosstalk.

These characteristics are similar to what a dedicated Audio Analyzer measures, the difference being that the processing is done by the PC’s processor in RMAA and the dedicated hardware in an Audio Analyzer.

Note
Even between Audio Analyzers manufactured by the same company, the precision is often different between models - this is caused by the sensitivity of the internal equipment, just as the RMAA is dependent on the performance of the ADC inside the computer. Simply put, you cannot measure half an inch (small distortions) with a yardstick (clumsy equipment). This is because the ADC and the Analyzer themselves have a noise ceiling.

The range over which RMAA can measure signals depends on the setting of the sound card used. In case of regular studio equipments, 0dB is calibrated to be +4dBU(=1.228V = +1.78dBV), and for less powerful consumer devices 0dB is calibrated at -10dBV(=-7.78dBu = 0.316V).

The problem is that often, cellular phones or portable audio players aren’t even able to reach the -10 dBv mark. In cases like these, there are two options: the first is to commence with testing using an external amplifier to set the volume to 0 dB; the other option is to commence with testing as is with signals failing to reach a sufficient level.

The former offers a convenient way of testing, but the results may be affected by the performance of the amplifier used, and even worse, to varying degrees (as each device has different output levels) - this significantly undermines the accuracy of this method.

Thus, in such a situation, Golden Ears opts to commence testing as is, without further amplification, instead relying on increased sensitivity to noise (and thus a lower observable noise ceiling) by testing in 24-bit mode.

 
The output signal is usually connected to other devices (such as loudspeakers) - in most cases, the input impedance of the connected device is sufficiently large that there is no meaningful discrepancy in the actual sound quality and the quality measured in RMAA under a loadless test.

However, when the load only has a small impedance - such as in case of 1/8’’ jacks used with earphones, or small portable speakers - the loadless test is not a good indicator of what you will actually hear. The reason for this is because lower impedance draws an inversely large current - or is supposed to. In real life, often the device cannot properly supply this current due to various reasons (maximum allowed current transistors in the OP AMP, or the limit of the power supply).

Thus, the sound might clip or distort with a minute increase in volume when the device in question is paired with a small impedance, which cannot be detected with loadless testing alone - it is for this reason that Golden Ears conducts loaded tests in addition to the standard loadless test when necessary, with speakers or earphones plugged into the device being tested.