computes the loudness of audio according to the EBU momentary definition.
computes the loudness according to the definition def.
Details and Options
- AudioLoudness tries to locally compute a measure of the perceptual amplitude.
- AudioLoudness returns a TimeSeries of the computed loudness measure for each partition.
- Available settings for loudness definition def include:
"EBUMomentary" K-filtering, 400ms partitions & 100ms offset (in LUFS) "EBUShortTerm" K-filtering, 3000ms partitions & 100ms offset (in LUFS) "VUMeter" no filtering, 300ms partitions & 100ms offset (in VU) "Peak" max value, 400ms partitions & 100ms offset (in dBFS) "TruePeak" max value, 400ms partitions & 100ms offset, resampling to 192 kHz (in dBFS)
- Loudness Units Full Scale (LUFS) is an absolute measure. The recommended target loudness level is usually LUFS. One unit of LUFS is equal to one dB. It is also known as LKFS (Loudness K-weighted Full Scale).
- K-filtering is the process of applying a filter designed to approximate human loudness perception. It is an integral part of the loudness definition by the EBU.
- Decibels relative to Full Scale (dBFS) can be used to measure amplitude or RMS levels. 0 dBFS corresponds to an amplitude or RMS value of 1.
- Volume Unit (VU) is a relative measure. One unit of VU corresponds to 1dB. The zero is placed at the level corresponding to dBFS.
- For multichannel audio signals, loudness is computed on the average channel values.
- The following options can be specified:
Alignment Center alignment of the time stamps with partitions Padding Automatic padding scheme PaddingSize Automatic amount of padding MetaInformation None include additional meta-information ResamplingMethod Automatic the method to use for resampling paths
Examplesopen allclose all
Basic Examples (1)
The "EBUMomentary" and "EBUShortTerm" definitions differ only in the partition size:
"EBUShortTerm" uses larger partitions:
The "VUMeter" definition has very similar characteristics to the "EBUMomentary", but no K-filtering is performed:
Use the "Peak" or "TruePeak" definition to check if the signal clips at any point:
Use the Alignment option to control the position of the time stamps of the result:
Use the MetaInformation option to add meta-information to the result:
Use the Padding option to control the padding scheme:
Use the PaddingSize option to control the amount of padding:
Compute the average loudness of a recording:
Compute the loudness range of a recording:
Perform a time-dependent loudness correction:
Compute the time-dependent amplitude ratio of the signal to the target loudness:
Multiply the original signal by the ratio to get an approximately constant loudness result:
Divide the original signal by the ratio to expand the dynamic range of the signal:
Properties & Relations (1)
The "TruePeak" definition performs a resampling that simulates a digital to analog conversion:
In the case of a square wave all sample values are within -1 and 1:
Whenever this signal reaches a digital to analog converter (such as a sound card) an interpolation is done. The "TruePeak" algorithm simulates this operation using a 4× resampling:
Even though all sample values were within acceptable limits, the resampled signal shows clipping:
This is the reason for the discrepancy between the "Peak" and "TruePeak" algorithms.
Wolfram Research (2017), AudioLoudness, Wolfram Language function, https://reference.wolfram.com/language/ref/AudioLoudness.html.
Wolfram Language. 2017. "AudioLoudness." Wolfram Language & System Documentation Center. Wolfram Research. https://reference.wolfram.com/language/ref/AudioLoudness.html.
Wolfram Language. (2017). AudioLoudness. Wolfram Language & System Documentation Center. Retrieved from https://reference.wolfram.com/language/ref/AudioLoudness.html