This node constructs a falloff from the input sound. The nature of the falloff depends on the mode of operation of the node. All mode of operations have a number of common inputs and options that we shall introduce in the following sections.

**Sound**- The input sound.**Frame**- The input frame to analyse the sound at.**Attack**- The*Attack Time*, a value that defines how fast the sound intensity increases. A low value means the sound intensity will rapidly increase, while a high value means the sound intensity will slowly and gradually increase. This value only have an effect if the*Smoothing Samples*in the*Advancted Node Settings*is not zero.**Release**- The*Release Time*, a value that defines how fast the sound intensity decrease. A low value means the sound intensity will rapidly decrease, while a high value means the sound intensity will slowly and gradually decrease. This value only have an effect if the*Smoothing Samples*in the*Advancted Node Settings*is not zero.**Amplitude**- The maximum value the sound intensity can reach.**Low**- The lowest sound frequency to be considered. A value of zero means the lowest possible frequency while a value of one means the highest possible frequency.**High**- The highest sound frequency to be considered. A value of zero means the lowest possible frequency while a value of one means the highest possible frequency.**Scene**- The target scene. This only affects the frame rate of the animation.

**Falloff**- The output falloff.

**Fade To Zero**- The node will consider the intensity of the terminal frequencies to be zero.**Low Frequencies**- The node will consider the intensity of the low requency to be zero.**High Frequencies**- The node will consider the intensity of the high requency to be zero.

**Reduction Function**- To compute the intensity of the sound at a particular frequency, a range of neighbouring frequencies are sampled and are*reduced*to a single value through a reduction function. The possible reduction functions are as follows.**Max**- The maximum of the frequencies is used.**Mean**- The average of the frequencies is used.

**Smoothing Samples**- The number of samples used to compute the intensities. Multiple samples are needed to achieve attack and release times. A high values results in a more accurate and smoother result but takes more time to compute.**Kaiser Beta**- Beta parameter of the*Kasier*window function. High values corresponds to higher main-lob leaking and lower side-lobe leaking. If you are not sure what that means, leave the value at 6.

In this mode of operation, the output falloff will be an index-based falloff
where the falloff value of an element is equal to the average spectral
intensity between the input low and high frequencies at a certain time. The
element at index 0 will have the average value at the input frame, the element
at index 1 will have the average value of the frame preceding the input frame
by a certain interval, and so on. See the *scale* input below for more
information about the interval.

One can think of this mode of operation as an average intensity with a temporal trail. As the index increase, the value is the average intensity at some previous frame. Consequently, this mode of operation can be rather expensive for any moderately large evaluation domain, since many spectral analyses has to be performed in a single execution.

**Scale**- The scale defines the length of the interval at which values are sampled. For an input scale`s`

and an input frame`t`

, the values will be sampled at`t`

,`t - s`

,`t - 2s`

, and so on.

In this mode of operation, the output falloff will be an index-based falloff where the falloff value of an element is equal to the spectral intensity at a certain frequency. The node only computes a limited number of spectral intensities, called the frequency bins. The frequency bins will be distributed along the input distribution in the frequency domain specified by the input low and high values. See the inputs in the following section for more information.

**Extension Type**- The number of elements that will be assigned to the computed frequency is limited and is controlled by the*Length*input. Other elements will be given values by extending the range somehow. The extension mechanisms available are listed as follows.**Loop**- The frequency bins will be repeated. For instance, if the input*length*is set to 10, elements with indices from 0 to 9 will be assigned the computed frequency bins from the lowest frequency to the highest in order, elements with indices from 10 to 19 will be assigned the same freqiuency bins from the lowest frequency to the highest in order, and so on.**Mirror**- The frequency bins will be ping-pong repeated. For instance, if the input*length*is set to 10, elements with indices from 0 to 9 will be assigned the computed frequency bins from the lowest frequency to the highest in order, elements with indices from 10 to 19 will be assigned the same freqiuency bins but from the highest frequency to the lowest in order, and so on.**Extend Last**- The extra elements will be assigned the value of the last frequency bin.

**Count**- The number of frequency bins to compute.**Distribution**- An interpolation that defines the distribution of the frequency bins. Typically, an exponential distribution will be used to get more bins from the low frequency regions as opposed to the high frequency regions.**Length**- The number of elements that will be assigned the frequency bins. If more elements are evaluated, the frequency bins will be extended depending on the extension type explained above.**Offset**- An offset that is added to the index of the element before evaluating the falloff. The offset can be used in conjunction with extensions to move the start and end of the frequency bins.

In this mode of operation, the output falloff will be an index-based falloff
just like the *Spectrum Index* mode. The only difference is that the index of
the elements will be the value of the input falloff at the element. So, in
reality, it is a falloff-based falloff. For instance, if the input falloff is a
*Point Distance* falloff, the elements closer to the center will be assigned
the lowest frequencies while the elements further away will be assigned the
highest frequencies.

**Falloff**- The input falloff.