Impulse Response

How to select the appropriate IR

Impulse Response (IR) is a "footprint" of the acoustic characteristics of the given space (church). It may be used to reproduce these characteristics with the help of dedicated software (such as Pristine Space). Perhaps the most distinctive acoustic characteristics of the space is the reverberation time (i.e. how long the reverb is). Therefore, the IR´s on this web site are ordered according to the reverberation time (in seconds). The bigger this number, the longer the resulting digital reverb will be. However, there are other features according to which you might choose your favorite IR.

Acoustical Parameters (ISO 3382)

For each of our IR collections, we provide a selection of acoustical parameters according to the ISO standard 3382. A clever utility of Angelo Farina was used to measure these values. Probably the most interesting ones are various kinds of Reverberation Time (EDT, T20, T30) and Clarity C50 and C80.

RT, T20, T30 - Reverberation time (RT) is defined as the time needed for the sound to decay 60 dB below the original value. It is usually measured as the time for the sound to decay 20 dB which is then multiplied by 3, or the time to decay 30dB, then multiplied by 2. The reverberation time is frequency dependent. Every frequency band may decay differently. Usually, very high frequencies decay very soon, the maximum RT is somewhere between 500-2000 Hz. A simple graph featuring the frequency decay for various frequency bands is shown. Colors indicate the intensity of the sound: yellow - high energy, red - less energy, violet - low energy, black - no energy (background noise).

EDT - early decay time, the time needed for the sound to decay first 10dB, then multiplied by 6. The shorter EDT, the better for the clarity of music, however, this recommendation is made for concert halls, while it is difficult to say the same for churches and organ music.

Clarity (C50, C80) - ratio of the early sound energy (first 50 ms, resp. 80 ms) to the later reverberant energy. The lower the value, the more "muddy" sound. For concert halls, it is usually recommended that the values are kept in the range of +1dB to -4dB, but of course, this cannot be easily applied to church acoustics and organ music.

Please, bear in mind that our graphs and tables reflect only one channel (typically the left channel of the centre IR) of one sample IR from the given collection. The parameters of the other IR´s from the same collection may differ substantially from the presented one.

Elucidation of the file names of IR´s

The filenames of our IR´s may appear mysterious at the first sight. However, there are very simple conventions which allow easy orientation what the IR represents.

Microphone techniques

If the IR filename does not start with the name of the IR collection, then it starts with a word describing the microphone technique used to capture the IR:

• front - stays for the O.R.T.F. technique, the microphones being directed towards the source, good for the front channels of your audio
• rear - this is the IRT-cross rear channels (microphones directed away from the source), good for the surround channels of your audio
• omni - stays for the AB recording technique with two omni microphones spaced about 1.5 meters apart.
• jecklin - the recording technique using so called Jecklin disk.
• head - recording using dummy (or real) human head
• NOS - N.O.S. recording technique

Whenever there is a scheme accompanying the IR, pictograms may be used to identify the positions of the source and that of the mics:

	AB - 2 spaced omni´s
	O.R.T.F. or N.O.S or a part of an IRT-cross
	dummy head (binaural)
	jecklin disk
	speaker (sound source)

Position of the source

Since an organ is unusually extensive and spacious object, it cannot be reduced to a single-point instrument. Rather, it may be compared to whole orchestra where every instrument occupies its place in the space. To simplify the model, it seemed adequate to reduce these many single-point instruments to several "typical" locations, providing IR´s specific for these locations. If the user breaks up the dry sound into different channels, he will be able to apply various IR´s to various parts of the organ (such as divisions, for example) reproducing the spatial differences between these parts.

Hence, IR´s on this site are grouped in collections. They represent the same space but different locations of the source. There are letters in the filenames allowing to discern the location of the source:

• LL - very left (good for wider spaced organ divisions, such as pedal towers)
• L - left
• R - right
• RR - very right (good for wider spaced organ divisions, such as pedal towers)
• C - center
• CL - center, but shifted a bit left (about 0.8m) - where it was not possible to measure from the very center (good for Rückpositiv)
• CR - center, but shifted a bit right (about 0.8m) - where it was not possible to measure from the very center (good for Rückpositiv)

Other distinguishing marks

Various versions of the same IR may have other letters or numbers before or after the name (1, 2 and similar). These have no other function than to indicate another version of the IR - typically different take.

Sometimes, there is a number indicating the sample rate (44, 48, 96) or bitrate (16, 24, 32).

Direct spike, balancing

After much thought, we deliberately decided to remove the direct spike from all our impulses and to put all of them in good L-R balance. Therefore, so called "purist approach" is not possible with our IR´s. There were various reasons for this solution, however, the main one is the ease of use for the potential users. The IR is ready to use without additional need to adjust anything and the wet output may be mixed with a portion of the dry sound if desired without worry about the artifacts. Our user gains more freedom of use at the expense of exact reproduction of the measured reality.

The other reason for removing the direct spike was specifically connected with organ music. Having an IR with a direct spike, you retain precisely the position of the sound source in the space (with respect to the invariant position of the listener). This would give good results for a point-like instrument. In the case of organ, it is not possible since the organ is not a point-like instrument. The direct spike would give misleading result since it would force wrong localization to the wet sound. Moreover, the organ samples for which these IR´s were designed, are not entirely dry which precludes the purist approach anyway. When recording the organ samples, we decided again to reduce the spaciousness of the organ to several typical locations (recording just outside the organ case with many microphones - at least L, C, R for each division), so that the samples themselves contain some localization information. This roughly corresponds to the sound source positions of the measured IR´s. Nevertheless, we firmly believe that this approach gives an adequate approximation of the entire complexity of the organ sound radiation.

For all what was said previously, it seemed good idea to do without the direct spike. Instead, you may mix the dry sound with the wet to simulate the location for the given pipe. The advantage of this approach is the possibility of user-adjustment of the virtual location of the sound source.