BCST1101 Audio Measurement

Human Hearing Sensitivity

  • humans can discriminate very small changes in the intensity (amplitude) of quiet sounds, but are much less sensitive to changes in louder noises

Dynamic Range

  • defines as the difference between the smallest and the largest values

  • in audio/ sound we are discussing the range of amplitude or loudness of the sounds

Air Pressure and Sound

  • Air pressure at sea level is about 101,325 pascals (Pa)

  • or 1 Atmosphere

  • About 10332 kg per square meter

  • there’s a constant pressure

Ambient Air Pressure

  • that constant pressure is also known as the ambient air pressure

  • ambient:

    - existing or present on all sides

  • sound is result of changing air pressure

    • sound is not constant pressure

  • atmospheric pressure is relatively constant

    • sound is pressure changes relative to ambient air pressure

No sound doesn’t mean no pressure

  • silence ~101,325 Pa

  • Sound is made by pressure going higher and lower than ambient pressure

Air Pressure and Hearing

  • the faintest sound humans can hear are changes in pressure measuring about 0.00002 Pascals

  • The loudest sound we can tolerate is somewhere around 200 Pascals

Lowest threshold

  • Or put another way:

    • the faintest sound we can hear is 20 micro pascals

Human Hearing and Loudness

  • loudness is the perception of intensity

  • how strong does a certain sound feel

  • human perception of loudness is not linear

  • doubling the power of the sound wave does not double the loudness

Power makes it heard

  • the power intensity ration between the faintest audible sound and the loudest sound we humans can tolerate is one to one trillion

One trillion to One

  • one million millions

    • 1 followed by 12 zeros: 1,000,000,000,000

  • So the ratio of human hearing sensitivity to the power or sound. audio is 1:1,000,000,000,000

  • this is our “dynamic range”

Loudness vs Volume

  • loudness is perceived

    • its how intense we think the sound is

    • so loudness is what we sense, what gets interpreted by our brain

    • meaning loudness is a subjective experience

    • what’s Lous to one person may not be as loud to another

  • volume is measured

    • volume is the measurement of the power of the sound and/or audio

    • volume is therefore an objective experience

    • what’s high volume on one device is high volume on other devices

Logarithms and measurement

  • because of the non linear response to loudness we need a scale that is also non linear

  • we use a logarithmic scale

  • Logarithms

    • one that goes up in steps of ten times per step to measure the power of sound and audio

    • why logarithms?

      • compare to a linear scale loudness scale

      • with a logarithmic scale

Linear Measurement

  • this scale has one Pascal pressure unit for each unit of loudness

  • this is called a linear scale, one step up for each step across

problems with linear audio scales

  • Soft sound levels cluster at the bottom and are hard to differentiate from zero

logarithmic measurement

  • a logarithmic scales make the different loudness levels clearer and easier to separate from each other

  • each step up the scale of sound intensity is 10 times the energy or power of the previous step

Logarithms and Decibels

  • decibels are a logarithmic measurement

Measuring sound using logs

  • for example, we have an amplifier with unlimited power output and a knob where each number on the knob sounds twice as loud as the previous number

  • 3 is twice as loud as 2

  • 4 is twice as loud as 3

  • they ay 10 watts is coming from our amplifier with the volume knob set to 1

  • our sound intensity is at 1 on a scale of loudness

  • that takes 10w of power

  • now we want to increase the intensity of the sound

Logarithmic scales

  • so we have a scale that can make big jumps in power measurement in a small space

  • no amplifier on earth has the ability to deliver 10 times with each step up on the knob

  • each one step of the scale increases the amount by multiplying by some amount

  • whereas on a linear scale, each ones step of the scale increases the amount of adding 1

f-stops are logarithmic too!

  • example the f-stops in a camera are logarithmic since each full stop up increases light coming in by a factor of 2

  • f5.6 is four times the light of f2.8

  • f5.6 is 2 stops up from 2.8

Why use a logarithmic scale?

  • because of the high dynamic range of human hearing

  • if we used a linear scale we’d need a trillion units to measure sound levels

  • with a logarithmic scale we end up with about 10 units or so

Logarithmic scale in audio

  • the logarithmic scale in audio uses powers of 10

Logarithms and Audio

  • we use a logarithmic scale

  • decibels

  • that goes by be powers of ten, to measure the power of sound

  • each step up the scale is 10 times the energy of the previous step

  • but not 10 times as loud

Logs,Bels and Decibels

  • bels?

    • the original audio scale was the Bel

      • a decibels isn 1/10 of a bel

      • FYI bels are too big of a unit to be useful

      • each step covered way too much power

    • so that means to get a useful measurement in decibels we need to do one more bit of math

Cutting Logs down to size

  • up to now we’ve ben talking about a scale that goes up 10 times per single step

  • that’s too big to be useful

  • so we’ve divided up the steps into 10 smaller steps

  • the decibel scale

  • all human dynamic range is from 0-120 on the dB scale

  • all human dynamic range would be from 0 to 12 on this scale

what is a decibel?

  • a decibel (dB) is a unit for comparing the intensity of two different sounds

  • it is not a unit of absolute measurement

  • it’s a relative scale

    • because decibels are a comparison, we need a reference point to use them

    • for example, sound in air is measured against air pressure at sea level

    • so the reference point is air pressure

How the relationship works

  • the scale for measure is decibels

  • the reference point is air pressure

    • aka sound pressure level

  • so the unit of measurement is the dB split express as 28 dispel

some weirdeness about spl

  • 0 dBspl should be no change in air pressure from the ambient 101,325 pascals, right?

  • but since humans can't detect changes in pressure less than 0.00002 pascals

OdB split is actually 0.0002 Pascals

  • why?

  • Because dispel is measurement of sound, 0 dB only needs to be the lowest pressure that humans can detect

  • not absolutely zero air pressure

FYI

  • FYI 120Dbspl is the “threshold of pain” where the loudness of a sound can actually hurt

  • since the threshold of pain is based on loudness not volume it can vary depending on the listener

Decibel scales DbFS

  • the terms dB FS means decibels relative to fulls Cale. it is used for amplitude levels in digital systems with a maximum avaible peak level where 0 dB FS is assigned to the maximum level

  • a signal that reaches 50 percent of the maximum level would, should have a value of -6 dB FS

  • all peak measurements will be negative numbers

why dBFS?

  • digital audio compares sound to numbers (bits). more bits = higher but depth = more sounds

  • digital systems are limited by the bit depth. 0 dbfs is chosen as the absolute maximum measurement

  • any single that exceeds the maximum ceases to exist since there are 0 bits left to count the data

dBu and dBv analog scales

  • what is dBu?

    • a logarithmic voltage ratio with a reference voltage of 0.7746 volt = 0dBu

  • what is dBv?

    • a logarithmic voltage ratio with a reference voltage of 1.00000 volt = 0dBv

  • analog audio compared sound to voltage and current

You cannot convert digital to analog

  • you can never match dBFs and dBu/dBv

  • dBu/dBv is volts - you measure it with a volt meter

    • analog audio: positive and negative voltage

  • dbfs is in contrast a binary number

    • digital audio: seroes and ones

  • the reference point of analog and digital are so different that they can’t be converted

  • instead you measure each as you do the conversion

  • example - send tone at 0dbu from the analog source. set the dbfs scale to -20dbfs to match the tone and start your transfer

Remember this power ratio

  • the original scale was Bels - 1 bel = 10 times the power, 2 bels = 100 times the power

  • divided bels by 10 to make a more useful scale, decibels

  • for decibels = 10 decibels is 10 times the power, 20 dB is 100 times the power, 30 dB is 1000 times more power

Sound and Audio

Amplitude

  • the distance above and below the centreline of a waveform

Frequency

  • the rate at which a sound wave repeats any number of cycles within one second

Wavelength

  • the distance between peaks of a wave in one cycle

Phase

  • the time relationship between multiple sound waves

  • measured in degrees

  • range 0-359

Frequency Range

  • the lowest and highest frequencies of a signal

Hertz

  • a frequency measurement unit (cycles per second)

Transient

  • a short duration, high level sound. Such as hand clap or snare drum hit

Axis - Microphones

  • on axis - if you envision a polar xis around a microphone, on axis would be at zero degrees

    • at the front of a mic

    • this is the area where the microphone’s frequency response is best

  • off-axis is at the side or back of the microphone

    • sound is distant, “thinner”, more “roomy”

Dynamic Range

  • the lowest and highest possible volumes of a signal

  • often used to describe limitation of equipment

Headroom

  • maximum amount of level equipment can take before you cause distortion by “clipping”

Noise Floor

  • the level of noise below which the signal is weaker than the noise

Signal to Noise ratio

  • this is the relationship between the signal and the inherent noise of the environment or electrical noise from equipment

  • measured between recorded sound and the top of the noise floor

Gain

  • gain - increasing the signal level

    • usually by amplification - adding electrons (increasing current)

  • unity gain

    • not adding or subtracting to the overall signal level

    • audio going in comes out the same level

Audio Compression

  • different from data compression, data compression is to make files smaller

  • audio compression is to increase the headroom without reducing the apparent volume

    • also to prevent slipping distortion

Treshold

  • the threshold is the level at which the compression effect is engaged

  • for example, if the threshold level is set at -10dB, only signal peaks that extend above that level will be compressed

Knee

  • the knee referee to how the compressor transitions between non-compressed states. Most compressors allow you to choose either a “soft” or a “hard” knee

Ratio

  • ratio specifics the amount of compression applied to the signal. this setting is expressed in decibels.

  • a ratio of around 3:1 is considered moderate compression

  • 5:1 would be considered medium compression

  • 8:1 starts getting into strong compression

Audio Distortion

  • a distortion is the alteration of the original shape of an object, image, sound, waveform other form of information or representation

  • in audio, distortion is the alteration of the shape of the sound or audio wave. this changes the sound of the wave

  • often but not always distortion is caused by having the signal at too high a level

    • this is also called overdriving, or over modulation or clipping

  • exceeding the headroom of your equipment is the most common cause of distotion

  • other causes exist

Distortion caused by clipping

  • abrupt limit to level, ie digital = hard clipping

  • flexible limit to level, ie analog = soft clipping

noise isn’t distortion, but distortion is noise

  • the addition of noise or other extraneous signals is not considered to be distortion

  • however, the effects of distortion are sometimes considered noise

  • any time an audio signal is “processed” some amount of distortion is introduced

Normal recording - no distortion

  • amplitude of the signal falls comfortable within the range

  • this is a well recorded signal

boosted signal - some distortion

  • here, the signal is amplified by 250%

  • recording components can no longer accommodate the dynamic range

  • strongest portions of the signal are cut off

  • this is where distortion occurs