Quality Parameters of Audio Signals
Quality Parameters of Audio Signals
 |
Quality Parameter of Audio Signals (530KB) Download this document in Acrobat format for printing and offline viewing. |
Sometimes signal level is not considered to be a quality parameter. But we have to realize, that for the listener the quality of the reception will depend strongly on the proper setting of the levels in the studio and at the transmitter.
Therefore proper levels are the first requirement for good sound quality.
The signal level is the only quality parameter which can be controlled during operation in the studio. When being trained in studio operation, the proper control of the level is the most essential skill.
2.1. The Audible Effects of the Signal Level
The correct signal level is to be controlled during mixing, recording, play-back and transmission of radio productions.
During any of these processes the following faults can occur:
- The relation of levels of two or more signals being mixed are not set correctly, so that the signals do not appear at the proper relationship.
- Levels between two takes or between music and announcement are not set properly, so that the listener experiences an unpleasant change in volume.
- An entire program or recording is set to a level too high or too low, so that the technical limits of the equipment are exceeded.
Furthermore the signal level has an effect on other quality parameters, namely the signal to noise ratio and the distortions.
- If the signal level is too low, the signal to noise ratio will be reduced,
- If the signal level is too high, the distortions will increase.
The reasons for these effects will be discussed in later chapters.
It is especially important, that the level must not only be correct at the output of the studio, but must be properly set for each stage of the signal chain. Therefore correct setting of levels of each equipment is essential.
At the transmitter the correct setting of levels has even further aspects:
- Setting levels too low, i.e. under-modulating the transmitter, is a considerable wast of energy, as it causes a reduction of service area.
- Setting levels too high, i.e. over-modulating the transmitter, leads to clipping and distortion or even tripping of AM-transmitters.
FM-transmitters will produced radiations which do not agree with the radio regulations.
The human ear is an unreliable instrument to judge correct signal levels.
- First of all it does not have the capability for absolute detection of sound level. It is impossible for somebody to tell if two signal have the same level, if they do not occur immediately after each other.
- The human ear is not "calibrated". We cannot tell the signal level just by listening, nor can we tell the level difference between two signals, occurring at the same time or after each other.
- The human ear is relatively insensitive to changes in signal level. A change of approximately 10% is necessary to detect an instantaneous change in level.
From this we can deduct two conclusions:
- We need an objective means to measure the signal level. This will be the program meter for the studio operator or the level meter for the maintenance technician.
- Audio signals need not to be measured with high precision.
A precision of ±10% is sufficient.
The technical quantities used to describe signal level take account of these facts.
2.2. Different ways to define signal levels.
Acoustical signals are difficult to handle. Therefore the sound is converted as soon as possible to electrical signals by the microphone. Signals in the studios are therefore electrical signals. These can then be amplified, recorded and transmitted.
Electrical signals have to be described by electrical quantities. Audio signals are always alternating signals, this means varying signals. We have to apply the special theories and mathematics which are necessary to describe these signals. Basically the audio signals can be described by their effective or RMS voltage and their frequency.
The typical signals in the studio, produced by music or speech, are not consisting of one amplitude and one frequency, but of constantly varying values and even of many voltages and frequencies at the same time.
For testing the equipment and getting a clear and definite result, it is necessary to use a signal having one constant amplitude and frequency. This results in the definition of test signals.
Examples of test signals:
U = 1.55V, 1kHz or U = 0.1V, 3150Hz
The adaptability of the human ear to different sound levels and the wide range of levels makes it useful to have a logarithmic scale for signal levels. As in the studio the signals are available as electrical signals, a relationship has to be defined between signal voltage and signal level. This is done by defining the DECIBEL (dB). The dB is a relative quantity with logarithmic characteristic. Depending on the reference quantity, many different forms of the dB are in use.
An additional letter is used to define which reference signal is used.
In audio electronics we mainly use
dB, dBr and dBu (formerly dBm).
The dB without any defined reference is just used to describe the relationship of two signals. It can not be associated with a certain signal level or voltage.
dB is used to express gain or attenuation:
dBr is used to express absolute signal levels relative to a given standard.
The standard must be defined somewhere or must be a common agreement.
(e.g. the standard studio level).
Often the dBr is simply called dB. (e.g. "This recording is 10dB too low.") Strictly speaking this is not correct.
dBu is used to express absolute signal levels. The reference used is an absolute voltage. The value of this reference voltage is defined historically.
It is the voltage which will produce a power of 1mW in a 600
resistor.
(Formerly the term "dBm" was used for this purpose.)
Any signal level can then be determined by the following expression:
Thus
0dBu = 0.775V
Any other signal voltage can be calculated, e.g.:
-4dBu=0.49V ; -2dBu=0.62V ; +6dBu=1.55V ; +8dbu=1.95V ; +20dBu=15.5V
Level meters used for maintenance and calibration of audio equipment are normally calibrated in dBu and volts or millivolts. Either scales can be used for measurement purposes, but in broadcasting electronics it is more common to use the dBu-scale.
The signal level needs not to be measured with high precision. Therefore it is normally sufficient to measure with a precision of 1dB, which means a tolerance of approximately 10%. Therefore also calculations can be done in a rather rough way. Often the use of the formulas, which would require the use of a calculator, can be avoided, if a few common relationships are memorised.
The following table shows the relationship between the most important values for signal ratios and dB:
With these values most dB-values or ratios can be converted by simple calculation.
Examples:
From dB to ratio:
64dB = 60dB + 10dB - 6dB 
1000*3/2 = 1500 (precise: 1585)
5dB = 17dB - 6dB - 6dB (7/2)/2 = 1.75 (precise: 1.78)
or
5dB = 6dB - 1dB 2 minus 10% = 2 - 0.2 = 1.8
from ratio to dB:
ratio is 50 
7 * 7 17dB + 17dB = 34dB (precise: 34dB)
or 50 = 100/2 40dB - 6dB = 34dB
ratio is 18 = 2 * 3 * 3 6dB + 10dB + 10dB = 26dB (precise: 25.1dB)
ratio is 1.6 = 16/10 = 2*2*2*2/10 4*6dB - 20dB = 4dB (precise: 4.1dB)
The Studio Level
When talking of levels during production and transmission of programs, the "normal" or maximum level is referred to 100% or 0dB. This is a relative level. It is therefore necessary to define the reference or the standard studio level. This will then be an absolute level (in dBu) or voltage. Of cause, during normal operation of the studio, it is not necessary to know the standard studio level.
The maintenance technicians have to take care, that all equipment is aligned to this level.
In Germany the standard studio level is +6dBu = 1.55V. Other countries have other levels between +4dBu and +12dBu. The actual value is not important, but
It is of absolute importance, that there is only one standard studio level for the entire broadcasting house.
Of course it is useful to adopt to some national or even international standard in this respect, and it is useful also for the non-technicians to know this value.
2.3. Dynamic Range.
The normal programme material in broadcasting produces constant varying levels. This means a range has to be defined, in which the signal is allowed to vary. This is called the dynamic range.
The dynamic range is limited towards the high levels and towards the low levels by restrictions given by the equipment. High signal levels are limited by the maximum signal amplitudes the equipment can handle. Exceeding this limit results in clipping and distortion of the signal. Low signal level should still stand out against the physical noise produced by all equipment. Therefore the "noise level" of the equipment is the lower limit for the signal.
Fig. 2.3.1.
Graphical representation of the dynamic range of a tape machine. The normal maximum signal is the standard studio level (0dB).
The head room is used to prevent accidental over-modulation, which leads to clipping.
The foot room is the required level difference between the lowest possible level and the noise level. It is required to prevent the noise from disturbing low volume passages of the programme.
The normal dynamic range of broadcasting programme material should not exceed 40dB.
During production and transmission constant control of the signal level is essential at all stages, to keep the signals in the optimum range of the equipment. Especially over-modulation produces very nasty effects and is annoying for the listener.
2.4. Programme Meters (level Meters)
Different tasks are demanded for level meters and different solutions have been developed to meet them; but although a meter cannot be used to judge such things as the balance of one voice against another, or speech against music, it is nevertheless an essential item of studio equipment.
The demands for the use of programme meters can be devided in two groups:
1. The objective demands to make optimum use of the equipment:
2. The subjective demands for good listening:
There are several types of meters which can be used to line up equipment or check for over-modulation:
- the peak-meter or peak programme meter (PPM) (used almost all over Europe)
- the VU-meter (used in America)
The peak-meter
One should correctly talk of a quasi-peak meter. It consists of a full wave rectifier circuit followed by a logarithmic amplifier and an indicating device, calibrated in rms-values.
Peak reading means: a 10ms full level sine wave results in 90% (-1dB) indication of a constant tone of equal amplitude. A pulse of 3ms results in an indication of
-4dB. This is the attack time or integration time.
The recovery time or fall-back time from 0 dB to -20 dB (or from 100% to 10%) is about 1.5s to 2s.
The scales of the different indicating instruments (meters) are almost logarithmic to facilitate the reading, since one wants to stretch the upper part in the total range of -50dB to +5dB, (or 0,3% to 180%, corresponding to an amplitude ratio of
1 to 600). In order to recognize better the overload limit during operation, the range is stretched between -6dB and +5dB, corresponding to 50% till 180%.
The accuracy of the indication is ±1dB between -5db and +5dB
and ±2dB between -40db and -5dB.
Fig. 2.4.1.
The characteristic between signal level and indication of a peak meter.
For the indication, different types of instruments can be used:
- Light spot meter,
in former years one of the most commonly used instruments.
- Bar graph meter
with neon discharge elements. Two independent measuring amplifiers and indicating bars make it useful for stereo. The bars are composed of up to 100 segments each, which gives the impression of continuous bars. The overload range (0 to +5dB) will illuminate at higher intensity. At present it is one of the most common peak meters.
- LED-meter
with up to 70 LED's. Amplifier and display are in one unit.
- Pointer instrument
with measuring amplifier and moving coil instrument in one unit. Often the instruments has its mechanical zero at the 0dB-point or at the right end of the scale.
Different types of Peak Programme Meters:
The display of a bar-graph PPM with discharge display.
A bar-graph PPM using LED.
A pointer type PPM with amplifier.
Requirements for Peak Program Meters
german versions:
Scale: -50dB to +5dB
rise time (integration time) to -1dB (90%) reading: 10 ms
fall-back time (return time) after 100 % reading: 1,5 s to -20 dB (10%)
2,5 s to -40 dB (1 %)
overshoot: +0,3 dB
Peak program meter - BBC-specification
The BBC version of the peak program meter is a pointer instrument. The scale has divisions from 1 to 7. Each division represents 4dB. The standard level (100 % modulation) is +8dBu at reading 6. A pulse of 4ms produces an indication of 80% of the full deflection. Fall-back time is about 8,7 dB/second.
The VU-meter:
A simpler type of meter than the PPM is the American VU-meter. It is standardized since more than 40 years in the USA and is composed of a very sensitive moving coil instrument with full wave rectifier and preceding damping network.
Specifications:
A 300ms voltage pulse results in a 90 % indication. The recovery or fall-back time and attack or integration time are equally 300ms. The meter has two scales, one is calibrated in decibels from -20VU to +3VU and is about logarithmically linear (more than half of the scale is occupied by a range of 3dB on either side of the nominal 100% modulation). Since little program material will remain consistently within such narrow range, the needle is generally either registering only small deflections or flickering bewilderingly over the full range of the scale. This scale is intended for use with steady tone for line-up. The other scale is calibrated in volume units (VU), numbered from 0 to 100 over about two thirds of its range, with the last part uncalibrated and marked in red. This scale represents percentage of full modulation and is intended for use on programme. A damping network increases the input resistance to 7,5 kOhm.
Different programme material lead to different indication behaviour, which demands some experience for exact control. Although the VU-meter is cheaper and lighter and needs neither power supply nor maintenance, it's disadvantages, compared to the PPM, cannot be neglected:
- low sensitivity,
- limited range,
- long attack time which underrates short peaks.
VU-meters require a so-called lead. This means the VU-meter is calibrated to indicate higher signal levels than there really are, to compensated for its impossibility to react instantaneously on varying levels of normal programme material. Different programme material (speech, classical music, pop music) will require different lead, ranging from 4dB to 10dB. Some VU-meters allow to adjust the lead according to the requirement of the programme. It is to be set according to the experience of the operator.
Most VU-meters have fixed lead settings, e.g. 6dB.
When constant test tones are applied to VU-meters, they will indicate some value which is higher than the true signal level (higher by the lead). Therefore
VU-meters can not be used to do measurements
or service alignments in the studio.
Fig. 2.4.6.
Scales of different VU-meters.
The functions and characteristics of the two forms of meters are very different and rather than one being better than the other, they complement one another very nicely. Indeed, a good engineer should be conversant with both types of meters and able to use them to advantage in order to get the most out of his equipment.
The VU meter displays a value which is more or less proportional to the loudness of a sound as perceived by the human ear. It is thus a kind of loudness indicator. The advantage of using a VU meter for mixing audio signals is that it gives a visual indication of the acoustic conditions being experienced by the listener. However, audio signals are neither symmetrical nor periodic (over a short period of time) and thus a VU meter is incapable of delivering reliable information as to the exact signal level present (volts or dBu).
In contrast, the PPM is a more scientific instrument which is basically accurate when it displays the peak level of a signal over a given period of time (200ms). It's rise time of 10 ms is short enough to ensure the correct measurement of most transients. The main advantage of the PPM is that it allows the operator - in recording or broadcasting - to obtain the maximum performance (in terms of signal to noise ratio and headroom) of his equipment. The principle drawback of the PPM is that the indication has little or nothing to do with the audio loudness as experienced by the listener.
2.5. Psycho-acoustical effects of the human ear.
As stated before, the human ear is not a calibrated measuring device. There are a number of physiological characteristics, which have to be considered. Some of them will be discussed here.
Equal Loudness Contours
The human ear will not recognize the sound levels of different signals equally. Experiments have shown, that frequencies around 4kHz are recognized with a much higher intensity, than very high and very low frequencies.
Fig. 2.5.1.
The typical curves of equal intensity for different frequencies of the human ear.
The curves will differ for different sound intensity. At low intensity the intensity differences are very big, at very high levels there are very little differences.
This relationship will be of some importance when discussing unwanted signals (noise).
Amplitudes of Sound
If we measure the amplitudes of all signal frequencies of normal sound (program material) we will find, that the different frequencies will occur at different level. As frequencies and levels are constantly changing, it is necessary to make statistical evaluations for different sound (music, speech, natural sound). The following curves show the resulting amplitude statistics.
Fig. 2.5.2.
A curve showing the amplitude statistics of natural sound. Note that these are statistical values and it will well occur, that some sound has a different amplitude distribution.
The curve shows that the amplitudes at frequencies below 200Hz and above 1kHz are getting less. At the ends of the audio band they are 20dB less than in the centre.
For the audio equipment this means, that some of the signal frequencies require less amplitude or power than the centre frequencies.
E.g. the tweeter of a loudspeaker normally has to handle less power than the medium frequency loudspeaker. (Therefore attention during tape cueing: Cueing produces an extraordinary amount of high frequencies, which can easily kill the tweeters of the monitor boxes. Reduce the volume during cueing!)
Masking Effect
The equal loudness contours are only relevant, if only ONE signal is investigated. If several signals are present at the same time, the stronger signals will cover or mask the weaker ones. This effect will be the stronger, the closer the frequencies of the two signals are together.
Fig. 2.5.3.
The curve for the audibility of a signal changes, if other signals are present, masking the signal under test. The closer the frequencies of the two signals are, the stronger is the masking of the weaker signal.
The masking effect is used in modern digital signal processing for the purpose of data reduction.
|
Quality Parameter of Audio Signals (530KB) Download this document in Acrobat format for printing and offline viewing. |
Quality Parameters of Audio Signals