Preserved Frequency Spectrum

The recordings of Munk Productions Ambisonic Ambiences contain the full spectrum frequency content as closely as possible when compared to the real world audio.


Low frequency

The MPAA files contain unattenuated low frequencies (LF) down to about 40 Hz and still contains significant LF energy at 20 Hz. This is not alway the case with ambience libraries, as we've discovered. Sometimes LF appears to have been rolled off or reduced. We don't do that. In fact, we don't change anything in the spectrum below 2 kHz (see also next section on high frequencies).

LF is crucial for envelopment and realism, as has been argued in later years (see Further Reading for more on that). Also, in the real world, most of the audio energy present lies in the lower frequencies. This is partly due to the fact that LF travels further in air than high frequencies (HF). Audio waves excite the movement of air molecules, which is how sound is transmitted to the ear. When doing so, energy is lost to friction (generating heat), and the faster these movements change direction, the more energy is lost.

So in many scenarios, LF dominates the audible sounds, and these frequencies must be present in the audio signal for it to sound realistic. This is clearly the case for nearly all outdoor recordings, but also for recordings of transportation methods like trains.


High frequency

Since we want to provide you with the most realistically sounding ambiences possible with the current technologies, we don't really do anything to the spectrum of the recordings.

And yet, there is one thing, we do do. We equalise binaural files. This is because of an inherent, minor HF loss when decoding Ambisonic files to binaural. The loss is highly dependant on the type of binaural decoder. We use IEM's BinauralDecoder (which we find to be the best sounding of the ones we've heard) and that has only a small 4-5 dB HF loss at 10 kHz, probably due to the HRTF set it is based on (for more on HRTFs and what they are, see Further Reading). The loss starts to appear from around 3 kHz and upwards. To mitigate this, we have done extensive listening tests and comparisons between binaural versions of recordings and real world sounds to find the equalisation parameters that result in the most realistic reproduction possible. These are implemented in the binaural files of MPAA.

If you do your own decoding, beware of this, so you don't get too dull sounding binaural files.