The History of Low-Level Audio Background Noise Measurement

This history focuses on a particular type of audio noise measurement.  This page does not concern high-level noise, such as the health-threatening environmental noise in workplaces, or in residential areas due to nearby road traffic or aircraft.  This history is of the techniques and technical standards for using electronic hardware - and now software - rather than trials with human listeners, to measure the degree to which humans are likely to be annoyed by low-level background noise while listening to something else, such as music, speech or any other sound signal of interest to them.  This includes especially when the audio programme material is of a low enough level that it does not mask the noise signal in the human auditory processing system. 

The German word Störgeräusch is used to refer to low-level background noise of various types, according to the degree to which humans find it annoying, disturbing or interfering.

This field primarily concerns assigning a single number to the measured noise signal in isolation, rather than multiple numbers or a spectrum of what the human auditory tract would perceive. Such numbers should enable an educated person to reliably evaluate, for instance, the self-noise of different microphones and/or pre-amps, in a way which matches the actual performance of these devices in typical recording environments.

This field is not directly concerned with the masking properties of the auditory tract (one high level signal making an otherwise audible lower level signal at a different frequency less audible).  this would be required to model the perceived degree of background noise interference in the presence of a given audio programme.

The most significant technical standards in this field are CCIR 468 and its derivatives, involving quasi-peak detection of the filtered noise signal.  However, these techniques are only used by some equipment manufacturers at present.  Many or most manufacturers give dBA figures for background noise.  So this history concerns debate between the relative merits of dBA measurements (filtering followed by RMS detection), the CCIR 468 approach and other approaches.

These techniques are of particular interest regarding:

They may also be of interest for any other source of noise which may be present in a system, in combination with any or all of the above:

These sources of unwanted signal are not caused by or related to the audio programme.  (If they were, then this would be a form of intermodulation "distortion", including perhaps how the input signal alters the behaviour of a poorly designed sigma-delta analogue to digital converter to produce "birdie" tones.) The sum of these noise signals constitutes a potentially annoying "background noise" which detracts from listening enjoyment and/or the ability to accurately perceive the audio programme material.  The goal of the techniques and standards which this history concerns is to use hardware or software to measure the sum of these noise signals and to predict the annoyance or other disturbances this would cause to a listener, without any assumptions being made about the programme material. 

Since human audio perception is strongly affected by level, when measuring a noise signal in analogue electronic form or in a digital datastream or file, it is necessary to relate this to a physical noise level presented to the listener's ears.  For instance, when specifying the self-noise of a microphone, the self-noise figure relates to the equivalent sound pressure level in air which would be required to produce such a noise signal in a noise-free microphone.  When measuring the self-noise of a pre-amp and/or ADC, such as in a digital audio recorder, the measured self noise can be combined with a microphone sensitivity to yield an equivalent sound pressure level for these sources of electronic noise.  Ideally, we would be able to combine this measurement figure with the self-noise figure of the microphone and produce a third figure which accurately predicts the degree of annoyance or disturbance of the mic and recorder in combination, referenced again to the sound pressure of noise which would have to occur in air to create the same level of disruption if it was fed to a noise-free microphone, pre-amp and ADC.  I am not sure if the current state of the art supports reliable addition or other methods of combining these noise figures.

These techniques and technical standards are of particular interest to these two Audio Engineering Society (http://www.aes.org/standards/about/) Working Groups:

Participation in these Working Groups is open to people who are not AES members.  This page http://www.aes.org/standards/development/membership.cfm links to a form for requesting participation in AES Standards Working Groups.  Each WG has a mailing list with a membership list and archives which are open to members of the WG.  In March 2012, measurement of low-level noise was being discussed on the SC-04-04 mailing list.

This field is also important to the IEC standard on microphones 60268-4, which as of the 4th edition (2010) adopts quasi-peak noise detection as the mandatory method of measuring the inherent self-noise of microphones:


Please let me know of any other organisations or WGs which have an interest in this subject.

While the material I have written below is copyright 2009 to 2013 Robin Whittle, please quote and use portions of it as you please, with attribution to this page http://realfield.com/anm/history/ .  Please let me know if you cite or quote this material.  This is a stable URL and I intend to maintain and ultimately improve this page in the future.

Below are one or more PDFs of original papers, some from 60 years ago.  I have not fussed over copyright for these important papers.  They report on work done by publicly funded organizations and so should not be locked up in paid-for-access repositories.  In one case so far I have also provided an English translation of such a paper, thinks to the kind efforts of someone else.  I claim no copyright over these documents.

>>>  Scope
>>>  Overview of audio standards organizations
>>>  Timeline
>>>  Origins of the CCIR 468 weighting curve and quasi-peak detector
>>>  Major standards
>>>  Viewpoints
>>>  Links
>>>  References
>>>  Update history

This page only concerns low-level noise as perceived by humans.  It concerns noise with physical, acoustic, origins and noise which is generated by electronic circuits.  The noise of interest here is always of a low level - so this is background noise, and perhaps distortion products, which are ideally below the program signal, including when the program is silent.  This discussion is not concerned with measuring noise at higher levels, such as in industrial settings, aircraft noise etc.  This discussion is primarily about random noise, potentially with spiky impulses etc. as distinct from noise which has strong tonal elements.

My particular interest is measuring the self-noise of microphones, but ideally, I think, the same system of weighting and metering noise should be just as good for measuring noise from pre-amplifiers, power amplifiers, ADCs, DACs, analogue recording systems, radio transmission systems etc. 

In particular, I am keen to identify one pair of filtering and metering techniques which will give a single figure which correlates well with subjective experience, without there first having to be any assumptions about the nature of the noise itself - for instance, whether it is "smooth", "impulsive" etc.  

The most prominent existing methods of measuring these types of audio noise are:

My view is that the CCIR-468 weighting filter and Quasi-Peak detector is the best of the widely recognised techniques.

I am not trying to document the Peak Programme Meter vs. VU meter debate, however this is a parallel and closely related debate which has, no-doubt, influenced opinions on how best to detect audio noise and generate a meter signal.

A detailed description of audio standards, particularly from the IEC, appears in #Weeks-1999 (in the Audio Engineer's Reference Book).  Another overview is: #Bohn-2000 Bewildering Wilderness - Navigating the complicated and frustrating world of audio standards,  by Dennis A. Bohn.

AES 

Audio Engineering Society http://www.aes.org/standards/ .  These working groups are open to anyone, whether an AES member or not, but only for the purposes of developing standards - not for reporting activities. 

I am an AES associate member and I paid extra for free access to the AES Library, which has the AES papers listed below.  I am also a member of two AES working groups: SC-04-04 (microphones) and SC-02-01 (digital audio measurement techniques).  Members of working groups have access to some current and past standards.

ASA

American Standards Association, now the American National Standards Association:  http://www.ansi.org .

BSI

British Standards Institution http://www.bsigroup.com . I have not found any BS (British Standards) which play a primary role in this field.  There are British Standards which are identical to, and therefore derivative from, IEC standards.  For instance "BS EN 60268-4: 2004" (microphones) is identical to "IEC 60268-4:2004", which is the 3rd edition.  Previously, the BS standards had different numbers, so "BS 6840-4:1997" was identical to "IEC 60268-4:1997" (2nd edition). 

CCIF

According to the ITU history , under the ITU (International Telegraph Union) CCIF (International Telephone Consultative Committee) was formed in 1924.  The next year, the CCIT (International Telegraph Consultative Committee) was formed.

CCIR

The ITU formed the International Radio Consultative Committee (CCIR) in 1927. (Another ITU history.)  Note:  the standards CCIR 468-1, 468-2 etc. are different versions of the one standard, as it was updated over the years.

CCITT

In 1956 the CCIT and CCIF merged to form International Telephone & Telegraph Consultative Committee (CCITT).

DIN

http://www.din.de  Deutsches Institut für Normung, the German Institute for Standardization has influence far beyond Germany.  For instance, in electrical connectors and standards for paper size (A4 etc.).     

IEC

International Electrotechnical Commission http://www.iec.ch was founded in 1906 and became part of ISO (International Standards Organization) in the 1947, ISO's foundation year.

IEC standard numbers had 60,000 added to them in 1967, so IEC 268-1 became IEC 6028-1.  The "-1" does not represent a version number, as it does with CCIR.  It denotes a particular standard within a larger set of standards.  See list at: en.wikipedia.org . 

So IEC 60268-1 is the first of a set of 18 standards concerning audio.  IEC 60268-4 is a separate standard, which happens to refer to some parts of IEC 60268-1.  Each such IEC standard has one or more versions denoted by an edition number.  For instance, IEC 60268-4 on microphones, is (2009) about to be updated to its 4th edition. 

ITU-R

http://www.itu.int/ITU-R/ In 1989,  the ITU was re-organized, ending the previous divisions of CCIR and CCITT.  (However #Weeks-1999 states that this change occurred in 1994.) These were replaced by ITU-T for Telecommunication Standardization, the ITU-R for Radiocommunication Standardization and the ITU-D for Telecommunication Development.  ITU-R standards may include "BS", such as "ITU/R BS.468-4".  I understand this stands for Broadcasting service for Sound.

Here is a 2002 version of Recommendation ITU-R BS.468-4 Measurement of audio-frequency noise voltage in sound broadcasting, of which there were previous versions in 1970, 1974, 1978, 1982 and 1986.  This describes a weighting curve and an inductor-capacitor circuit which realizes it.  ITU-R-REC-BS.468-4.pdf

Ulrich Steudel #Steudel-1933 performed work on the perceived loudness of impulse sounds, repetitive impulses and pure tones with various attack and decay characteristics.  This is cited as foundational work for this field in #Belger-1953.  Please see #Steudel-1933 for a graphic scan of this article, which I received in June 2014 and for the English translation, which I received in February 2016.

In June 2014 I was informed that an English translation of one sentence (around the bottom of page 123 is:


In February 2016 with the full English translation of this paper, we can see that this statement refers to Fig 18. and spans the end of page 123 and the start of page 124.  In the new translation, it is:


Here are the two signals:

The CCIF devised a noise weighting curve which was replaced in 1949.  The 1968 BBC report [#BBC-1968-EL-17 page 8] described it as:

See the 1952 section below for a depiction of this curve.

Impulsive Interference in A.M. and F.M. , BBC Research Department Report G.036, Serial 1947/12.  #BBC-1947-12.

There is a BBC Research Department report G.040 serial number 1948/24 Experimental Correlation Between Aural and Objective Parameters of Electrical Noise. This contains details of the weighting curve, meters and a photo of the experimental apparatus and listening room.  #BBC-1948-24 .

According to #Wilms-1970 (p 653), in 1949, the CCIF specified a weighting curve which was adopted in 1954 by DIN in DIN 45 405.   See the 1952 section below for a depiction of this curve.

This is the "P" curve as discussed below in the 1970 section and, according to #Wilms-1970 page 653 references 18 and 19 and page 656, reference 19, to the 1952 Mangold paper #Mangold-1952 ,  this curve was first suggested by AT&T.   The full quote from Wilms-1970 is in the 1970 section below.  If this is correct, then the AT&T suggestion must have been before 1949.

#Wilms-1970 page 652 states that this "P" curve was used by "CCIF 1949 and CCITT 1954", and the reference for both of these is the 1960 CCITT P33 Psophometer for Broadband Circuits. #CCITT-1960. Wilms also describes the response times of the "power proportional" (quadratic with sound pressure, that is voltage squared) metering circuit specified by this standard.

The BBC report [#BBC-1968-EL-17 page 7] cites what is apparently the same 1949 CCIF standard as #CCIF-1949 .  According to #Dolby-1978 the curve used by DIN 45405 in 1972 (the 1967 version) was "developed in the 1940s".

This "P" curve was used by DIN 45 405 until November 1983 (see 1997, Stephan Peus' paper) when it was replaced by the curve ("Q" in #Wilms-1970).  The "Q" curve is used by CCIR 468-x, now ITU-R 468-4 and by other standards which refer to these standards or reproduce their requirements.

#Belger-1953 also discusses the origins of this CCIF 1949 weighting curve.

According to #Smith-1970a there is a standard:

I have found various references to a psophometric weighting curve specified in "CCIF-1951".  I am not sure it dates from 1951 and I am not sure exactly what it covers.  There are various "glossary" descriptions such as http://www.atis.org/glossary/definition.aspx?id=2413 referring to "CCIF-1951" involving a reference of 800Hz rather than 1kHz.  For instance, page 1365 of the $1180 Computer Science and Communication Dictionary Vol 2: (Kluwer / Springer 2000, here on Google Books) has such text.

H. Mangold published a paper which I have not seen a copy of,  but would like to: #Mangold-1952.  Grundlagen der Geräuschspannungsmessung [Fundamentals of psophometric audio noise measurements], Rohde und Schwartz Mitt., no. 1, pp. 21-36 (1952).  This is cited by three papers mentioned here: #Belger-1953, Wilms-1970 and #Hertz-1977 . The latter does not make specific mention of the paper, but includes a graph of some weighing curves, from 1952.  Since the Mangold paper is the only reference dated 1952, I assume that the graph is from Mangold-1952:


"Vorschlag an CCIF" translates roughly as "suggestion on CCIF".

The solid line "CCIF 1949" curve seems to match the CCIF curve in the graph reproduced below from the 1968 BBC report.  The above graph does not show a curve equivalent to the "D.P.B" curve in the 1968 report, which was based on work by Ernst Belger published in 1953 #Belger-1953.  The D.P.B. curve evidently became the one chosen for CCIR 468. 

On this basis, we might conclude that the "D.P.B" curve, which seems to be the same as the Q curve in Wilms-1970, was not known to H. Mangold in 1952 but was published the next year.

Ernst Belger, of Institute fur Rundfunktechnik (Hamburg, Germany) published a paper #Belger-1953 E. Belger, Über die Messung und Bewertung yon Stögeräuschen [On the Measurement and Evaluation of Disturbing Noise].  In June 2013 I was delighted to receive an English translation of this important paper - follow the above link to the references section.

Belger establishes the importance of peak detection, with carefully chosen rise and fall times for the capacitor which is charged by the peak detector, together with the choice of a suitable weighting curve (audio filter).  It seems that these techniques later became known as "quasi-peak detection". 

This paper describes the history of the field, the experiments he and his colleagues conducted, and how they devised a weighting curve which was most suitable for the task of measuring the disturbing quality of background noise.  Later documents indicate that this weighting curve was adopted in CCIR 468 and so is used to this day (2013) in the standards which are based on CCIR 468.

According to #Wilms-1970, this paper, or at least work around this time by Ernst Belger (see mention below of his 1954 paper), appears to have had a lasting influence on the field regarding both weighting curves and quasi-peak detection.

In June 2013 a correspondent in Germany kindly provided the following note on the term  "Störgeraüschen", or more correctly  "Störgeraüsche":


The marvelous German way of building ever-more-specific terms by combining others also results in  Geräuschspannungsmessung (#Mangold-1952) which Google Translate recognizes as "noise measurement".

According to #Smith-1970a there is a standard:

According to #Wilms-1970 there is a standard:


which he links with "CCIF 1949".  According to #Wilms-1970 this was replaced in 1970 by CCIR 468.

(I suspect there is confusion between CCITT, CCIR and CCIF in accounts of these earlier standards.)

According to: http://en.wikipedia.org/wiki/Noise_measurement , CCITT P53 is now continued as CCITT 0.41.  On page 3-109 of the Audio Engineer's Reference (2nd ed. 1999) is the following description:



According to #Hertz-1977 (page 9):


In an earlier section, page 8, Hertz describes the Quasi Peak rectifier as being:


(See 1974 for further quotes from Hertz-1977.)

Ernst Belger's second major paper in the field appeared in 1954: #Belger-1954:

According to #Smith-1970a there is a standard:

According to #Langmuir-1962 the IEC published ASA S1.4-1961, which was a revision of ASA Z24.3-1944.  #Wilms-1970 gives the title for ASA S1.4-1961 as "Specification for General Purpose Sound Level Meters". 

#Wilms-1970 also states that IEC 268-1 (1968) uses ASA S1.4-1961 and IEC 179 (1965) as a specifications for its sound level meter.

#Langmuir-1962 has a graph and includes:

The paper #Langmuir-1962 describes (p 322) a "quasi-RMS" metering circuit with shaped magnetic poles, but this is not really RMS. 

A bridge rectifier charges a capacitor via a resistor r and the capacitor is discharged via a total resistance (r1 + r2), where r2 is the resistance of the meter.  A second capacitor can be placed in parallel with the meter for slower response.

#Peus-1997 mentions, without reference to any earlier version, that


Please see the 1997 section for the rest of this quote, with a diagram, explaining how DIN 45405 was changed in 1983 so the weighting curve was in line with CCIR 468.  The paper warns that this gives rise to two different technical specifications for DIN 45405 and that it is important to state which standard was used in a given measurement.

#Wilms-1970 (page 652) mentions IEC 179, 1st ed. "Precision Sound Level Meters" (1965) as one of the options for the meter (the other is ASA S1.4-1961) for IEC 268-1 (1968).  This page also implies that IEC 179 and the ASA standard both use "A" weighting.

#Hertz-1973 cites IEC 179 as:  "IEC Publication nr. 179 1965. Precision Sound Level Meters.

This was evidently extended with IEC 179A in 1973 to work better with impulsive sounds. 

#Wilms-1970 mentions a standard from the German Deutsches Institut für Normung,  (now http://www.din.de) DIN 45 405  (now referred to as DIN 45405):


Is this the first standard to specify noise detection with a quasi-peak detector?   This 1967-07-01 version was replaced in 1983-11-01.  In 2009-09, the 1983-11-01 version of DIN 45405 is still current and available for purchase.  It is in German and its title in English is: "Noise level measurement in sound systems".

See #Wilms-1970 for a more detailed description of this standard in 1970 and how it may have contributed to the adoption of quasi-peak metering in CCIR 468-2 in 1974.

DIN 45 633-2 is released (AKA DIN 45 633 part 2).  This is not to be confused with  separate standards 45 655-1, 45 633-3 etc.

There is very little trace of this standard in 2009, but it seems to have played an important role in this field.  The full details of the standard, taken from Wilms-1970 reference 9 are:


I suspect that IEC 179 or IEC 179A (some web pages indicate the latter is from 1973) is related to this DIN standard. They are all concerned with accurately measuring impulses of noise.  I think this is mainly safety and high levels of noise, where A weighting is more appropriate than it is for low levels of background noise. 

Some information on the response times of the metering systems described by DIN 45 633-1 and DIN 45 633-2 can be gleaned from:

http://www.rz.fh-muenchen.de/home/fb/fb05/akrt/Messtechnik/hp/kap5.pdf


A translation is:

#Wilms-1970 mentions IEC 268-1 1st ed, (1968) "Sound System Equipment" and describes it as having unweighted or A-weighted filtering and RMS detection (quadratic with sound pressure) with time-constants according to one of the two previously defined meter standards.

See the #Wilms-1970 for a more detailed description of this standard in 1970.

This 1st edition was replaced with a 2nd edition in 1985.  I don't have a copy of the 1st edition. The 2nd edition, 1985, with two amendments in 1988, is available to members of AES SC-04-04. I understand this is still (2009-09) current, although it is now known as IEC 60268-1.  IEC (60)268-1 is the first of a series of standards.  One of them - (60)268-4, concerns microphones and separately from (60)258-1 specifies weighting filters and a metering system.  IEC (60)268-4 is about to be revised to a 4th edition in late 2009 or in 2010.

#Wilms-1970 also mentions DIN 45 633 Part 2:


See 1970 below for a quote from #Wilms-1970 which indicates that  this standard too may have contributed to the adoption of quasi-peak detection in CCIR 468-2??? in 1974.

BBC report: The Assessment of Noise in Audio-Frequency Circuits

This report [#BBC-1968-EL-17] is freely available.  I think it is essential reading.  I understand it played in important role in the creation of CCIR 468-1 (1970)   Here are some excerpts:


See below ????????????? where  #Belger-1953 is cited as the original source of a curve which was later chosen as the CCIR 468 curve, the preferred weighting curve to this day (2009).

The report mentions there were four different types of meter and four different weighting networks either in use or proposed.  The BBC tests are described as taking place under conditions appropriate to VHF FM transmission. They restricted their research to existing meter and weighting systems, rather then developing new ones. 

Their tests used a high quality monitoring loudspeaker LS5/5, which is described at http://www.mhennessy3.f9.co.uk/rogers/bbc_txt.htm as a 3 way system with 12" woofer, 8" cone midrange and Celestion HF 1400 tweeter.  The speaker's effective bandwidth was 14 kHz and observers sat on the axis of the loudspeaker at a distance of 2.5 metres.  This involves less high frequency coupling to the ear than a side-on alignment, or the use of headphones. 

Subjects assessed noise in terms of the annoyance it caused and whether it "obtruded upon the observer's consciousness".  The results for these two criteria were closely matched.  The tests involved classical music and dramatic readings - not pop or jazz music or "didactic" speech such as news bulletins.

Ambient noise in the listening room was measured at 30 dB A weighted.  This corresponds to the quieter of the two listening rooms used by Belger [#Belger-1954].

The BBC tested four weighting curves.  One of them, the "D.B.P" (Deutsche Bundespost) is the "telecommunications authority in the Federal German Republic (West Germany).  This is the same curve later chosen for CCIR 468-1, and the quotation below gives some information about its earlier origins. 



The curves are described, with the first two descriptions paraphrasing:





A variety of meters were tested.  Two I will only mention: BBC Peak Programme Meter and the VU meter.  Those of interest are:





As a result of their tests, the BBC decided that the Modified OIRT meter was preferred.

They chose the DBP curve as being the best of those tested, with the OIRT curve being the next best choice.

CCIR 486-1

The first edition of the CCIR 468 standard was first published.  This is cited in #Wilms-1970 as:

???????????

While this standard was modified in some respects in later versions (468-1, 468-2, 468-3 and 468-4), its weighting curve ("Q" in the diagram below from Wilms-1970) has remained the same to the present day.  Quasi-peak detection was added in the 2nd edition CCIR 268-2 in 1974.  With CCIR's demise, the standard was passed on to CCITT and then to the ITU-R.  See quotes from Wilms-1970 below for more details.

Smith and Wittman, Shure Brothers Inc.

A. Douglas Smith and Paul H Wittman of Shure Brothers Inc USA publish Design Considerations of Low-Noise Audio Input Circuitry for a Professional Microphone Mixer  #Smith-1970a - a paper first presented in April 1969.  They mention three standards regarding noise measurement (see years above for more details):

• 1949: EIA (formerly RETMA) SE-101-A.
• 1953: IRE 19 S1 (now IEEE Standard 191).
• 1959: EIA RS-219.

This detailed 16 page paper presents a great deal of theoretical and practical information and states that noise must be measured with a true RMS meter ( p 144):

Herman A. O. Wilms' major paper

Herman A. O. Wilms published a paper Subjective or Psophometric Audio Noise Measurement: A Review of Standards #Wilms-1970 . 
I think this is essential reading for anyone concerned with the history of this field.

Wilms comments on #Smith-1970a, chiding its American authors for not mentioning European standards.   (They reply in #Smith-1970b .)

The origins of the term psophometer are given on page 651:


(The Ancient Greeks never fail to impress . . .)

He identifies three factors in the measurement process: the choice of a weighting filter, the choice of a detector and the choice of integration time.  The two types of detector are (page 651) :


Figure 1 shows the A weighting curve, the "P" curve of the 1954 CCITT DIN 45 405 (originally suggested to CCIF in 1952 by AT&T, as noted below), and the "Q" curve, as used to the present, which was used in CCIR 468, but which may have originated in #Belger-1953 ????.


Fig 1 from #Wilms-1970.

Other items from #Wilms-1970 are:

A list of international standards (see years above and notes below for more details) (pages 652 and 653):

• 1949 and 1954: CCIF 1949 and CCITT Recommendation P53 1954, though the reference is to a 1960 version of P53.
  
• 1961: ANSI S1.4-1961.
• 1965: IEC 179.
• 1967: DIN 45 405 (AKA DIN 45405.)
  
• 1968: IEC 268-1.
• 1970: CCIR 468, the first version AKA CCIR 468-1.
  

Here is Wilms' initial description of these standards.

IEC 268-1 (1968)

(Page 652.)


So the 1st edition of IEC 268-1, used A weighting with RMS metering.  We will see below that version 2 (1985) used the "Q" curve with quasi-peak metering (identical to CCIR 468-3).

IEC 268-1 (after 1997 IEC 60268-1) is a general standard, the first of a series.  It is not to be confused with IEC 268-4 (1st edition, 1972, now known as IEC 60268-4) in its 3 editions (a 4th edition in late 2009 or in 2010) which concerns microphones.

CCIF 1949 and CCITT 1954

I am not entirely clear what the CCIF 1949 standard was, but the "CCITT 1954" standard is cited as:


(Page 652.)


See 1951 for some notes on a "CCITT 1951" standard.

DIN 45 405 (1967)

(Page 652.)

CCIR 468 (1970)

The first edition was known as CCIR 468.  As far as I can tell, CCIR 468-1 was in 1974 and it was revised, but still known as CCIR 468-1 in 1976.  The original version from 1970 has the "Q" curve, but does not use quasi-peak metering. 

(Page 652)


Later editions of the same standard are known as CCIR 468-2, CCIR 468-3 and ITU-R 468-4.  All these later versions use quasi-peak metering and retain the "Q" curve of the first edition.

I think this reference to: "DIN 45 633/1 (impulse SLM 1968)" may mean DIN 45 655-2, which was released in 1968, rather than to DIN 45 633-1 itself.  So it seems that the two standards DIN 45 405 (1967) and DIN 45 633-1 and/or DIN 45 633-2 influenced the CCIR to adopt quasi-peak metering.


Fig 2 from #Wilms-1970.

Wilms on Quasi-Peak Detection

Wilms notes that the DIN 45 405 approach of "quasi-peak" is known as such because it involves a "fast but finite rise time specification".  This involves a faster rise time than 200 ms which is used in the other standards, including all the North American standards to date.  He notes that this 200 ms rise time resembles that of a VU meter.  (VU meters were preferred in North American while Peak Program Meters were preferred in Europe, including the UK.)  He also notes that the CCIR would probably adopt a faster rise time in the near future.  CCIR did so in 1974 with CCIR 468-2.

Here is what Wilms wrote on the history and implementation details of quasi-peak detectors for audio noise.  There are no references for the 1950 and 1962 dates in the first sentence.  The italics are in the original.  The following text appears to contain a recipe for constructing a quasi-peak meter according to DIN 45 633-2 (1968).

I don't yet have a clear understanding of DIN 45 633-1 or DIN 45 663-2, but I understand that there are multiple parts and that parts 1 and 2 are separate standards, or that -2 is an update which extends, but does not replace -1.  That is to say, they are not two editions of the one standard.  So:


A correspondent in Germany confirmed that the "-2" is an update to the original "-1":


(Pages 652  and 653.)


What was this 1962 standard?
 

Wilms on older standards for Weighting Curves

(Page 653.)

Wilms on Standards Current in 1970, including the 1953 origins of the "Q" weighting curve which is retained to the present

The "French Committee" (CCIR, I assume) objected to the use of A weighting in IEC 268-1.  The "1954 CCITT standard" was CCITT P53.


I assume the "German committee" refers to a DIN committee.  So it seems that Ernst Belger first devised the curve, which DIN suggested to the CCIR 398-1 study group - who recommended what we now know as CCIR 468-1.


In conclusion, he writes:

(Page 655.)

IEC 268-4 1st edition, on microphones, is released.  This uses A weighting and RMS metering for its noise measurements.   The 1st edition remained current until the 2nd edition was released in 1997.

Dolby Laboratories conduct a study on noise measurement, which is the basis for their proposal "CCIR-ARM" which was written about in their paper of 1978-79.  The study evidently resulted in Dolby producing a weighting filter box and this being widely used by the company's licensees.

I have found references to IEC 179A (IEC 179-A) which is described at http://www.city.vaughan.on.ca/..../Technical Definitions Noise Bylaw.pdf as a "first supplement to IEC 179", concerning impulsive sounds.  One article (Impulse Noise and Risk Criteria, J. Starke et. al. Noise and Health 2003, http://www.noiseandhealth.org/...aulast=Starck ) cites it as a 69 page standard "Recommendation for Impulse Sound Level Meters".  Please see 1968 for DIN 45 633-2, which I think is related to this.

Some details of IEC 179A's requirements, and of a corresponding DIN standard DIN 45 633-2 (year ???????) can be inferred from this brochure http://tubeaudio.ru/lab/2607.pdf for  B&K 2606, 2607 and 2708 precision sound level meters.

There are Fast and Slow IEC 179 responses and for impulse measurements, according to DIN 45633-2 (DIN 45 633 part 2 - 1968) and IEC 179 (on page 316):


So my impression is that the DIN 45633-2 standard of 1968 was effectively copied, followed or whatever by IEC 179A in 1973.

Bent F. Hertz presents a paper "A New Solid State Peak Programme Meter" #Hertz-1973.  This contains some history and technical details, including of a proposal for updating the 1965 standard  IEC publication number 179 "Precision Sound Level Meters".  This proposal presumably resulted in IEC 179A.

The references in this paper are not pointed to by specific citations in the text.  One of the references is a paper by Ernst Belger.  The journal details come from another source described below.


This translates as "Who measures correctly disturbing noise and pitch variations", for which the technical terms are "weighted signal to noise ratio" and "wow and flutter".    Googling the title turns up something in Google Books: a reference 207 on page 741 of Handbuch der Tonstudiotechnik by Johannes Webers. Edition 9, 2007 which gave me the journal details: Kinotechnik (1963), Heft 11, Seite  293. Googling reveals the full journal name is probably: "Verlag für Radio-Photo-Kinotechnik, G.m.b.H., Berlin-Borsigwalde, Germany". 

CCIR 468 amended twice as 468-1 for quasi-peak metering?

It seems there was an amendment to the original CCIR 468 standard in 1974 to specify quasi-peak metering.  No metering system was specified in the original.  See "1978" for more details.

#Hertz-1977 mentions in his bibliography two dates for "CCIR 468-1":


On page 9, following from what I quoted in the 1952 section, Hertz states:



I can't form a clear picture from this about the various forms of CCIR 486-1 in 1974 to 1976.


Hertz goes on to describe quantitization noise in ADCs and a technique for measuring it involving one or more "gliding sine waves".

CCIR 468-2 with quasi-peak metering  ?????

According to #Weeks-1999 in 1978 the second edition of CCIR 468 was released. Like the original, now known as CCIR 468-1, it has the weighting curve identified as Q in Wilms-1970.  In the first edition, no metering detector was specified. 

According to #Weeks-1999, the quasi-peak metering system was added in 1974, but this was presumably via an amendment, rather than a second edition. 

According to #Weeks-1999, in the second edition, a specific quasi-peak detector behaviour was "first completely specified".  This combination of weighting filter and quasi-peak detector remains current to the present day (2009-09), via later editions: CCIR 468-3, CCIR 468-4, now republished as ITU-R 468-4.

The ITU-R 468-4 document appears to be a copy of the 1986 CCIR 486-4 document, which contains a list of years, without further elaboration: "1970 - 1974 - 1978 - 1982 - 1986".  I assume that 1974 refers to an amendment.  The other years clearly refer to the original (-1) and the subsequent complete editions -2, -3 and -4.

Weeks states that these editions 468-2, 468-3 and 468-4 have not "introduced any significant modification to the instrument's performance" and that test equipment conforming to any of these three specifications "therefore may be assumed to give identical readings".  He also notes that this method "has been used in the UK since 1979 for most noise measurements on audio transmission systems by formal agreement between broadcasting and transmission authorities".  I understand this refers to long-distance audio transmission lines which carry radio program signals from studios to transmitter sites.

Dolby's "CCIR-ARM" paper

Dolby Laboratories presents a paper "CCIR/ARM: A Practical Noise-Measurement Method" at the May 1978 AES convention, which is published in its final form in JAES in March 1979 #Dolby-1978. 

This paper argued against the adoption of:

• Unweighed noise measurement of any kind.
   
  
• A weighting, with RMS detection, as used by NAB and in IEC 268-1 (1968).
    
  
• The weighing curve described by Wilms in 1970 as P which is used with a quasi-peak meter by DIN 45 405 (1967).
    
  
• The weighting curve described by Wilms in 1970 as Q and as used (??????? detector) in CCIR 468-1 (1970).

and instead proposed the adoption of a system they named "CCIR-AMR".  This uses the CCIR 468 weighting curve, but with a different reference level (effectively moved down by ~6dB) and with an average-reading voltmeter, rather than using RMS or quasi-peak detection.

The company had been using this approach to noise measurement internally and with its licensees, and had produced a small filter box which could be used with generally available average reading millivoltmeters.

The A weighting filter was rejected because it did not adequately represent the human sensitivity to higher frequency noise, which appears in analogue tape hiss.  Implicitly the deficiency is in the 1kHz or 2kHz to the 12kHz to 15kHz range.  The paper notes that measurements were frequently actually done with average weighting meters, in spite of the IEC 268-1 standard requiring RMS.

The DIN 45 405 approach was rejected on two grounds.  Firstly, the curve does not represent higher frequency noise adequately due to its sharp 9kHz cutoff.  The paper notes that this curve was developed in the 1940s for communication systems having sharp cutoff properties.  There is a note that the DIN 45 405 (1967) standard is "officially obsolete" - which looks wrong to me.  There is a note that a proposed standard DIN 45 500 (April 1975) uses A weighting, making it compatible with IEC 268-1.

The critique of DIN 45 405 included the first of several references to noise figures which were argued to be commercially unappealing or unacceptable due them being ~6dB or so worse than those resulting from the most widely used approach: A weighting with RMS detection.  The implication is that equipment manufacturers given the choice of several noise measurement techniques will tend to choose the one which provides the lowest number.


Dolby's critique of CCIR 468-1 and the conclusion of the critiques are quoted in full below:


Fig 1 of this paper depicts the A and DIN 45 405 (Wilms' P curve) together with a third curve representing the newly proposed (actually, from 1972) weighting curve: "CCIR/ARM":

Not shown is the true CCIR 468-1 curve (Q in #Wilms-1970 ) which is about 6dB higher than the so-called CCIR-ARM curve.   The 468-1 curve, as is usual for weighting curves, has its 0 dB reference at 1kHz.  The Dolby curve has its reference at 2kHz - apparently for the sole purpose of producing numerically lower results.

This paper does not explain why quasi-peak metering is expensive.  This expense purely affects test-equipment, which represents a very small part of the total cost of producing audio equipment, especially mass-market consumer hi-fi and professional audio equipment.

Dolby's criteria for choosing a noise-measurement system are quoted in full below:


I think few people would debate Dolby's arguments against no weighting, A weighting or the DIN 45 405 (P) curve.  The objections to the CCIR 468-1 approach can be summarized:

1. The quasi-peak metering system, although technically excellent, is prohibitively expensive and is (implicitly) not needed in professional or consumer audio testing, since it is really only needed when the noise contains significant impulsive elements.
    
   
2. The noise numbers produced by CCIR 468-1 are so much higher than those for A weighing and RMS detection that few companies would want to publish these figures.
   

The solution was to retain the CCIR 468 weighting curve, but to use essentially zero-cost average metering and then to ensure the figures the new system produces are not much higher than those resulting from A weighting and RMS detection.

Part of the reduction is due to the average meter producing a lower reading for a given random noise signal than a quasi-peak meter.  The average meter's output would also be lower than that of a true RMS reading meter.

The other part was achieved by simply jacking the curve downwards.  The formal arrangement was not an integer number of dB, but the 5.6 dB needed to make the curve read 0 dB at 2 kHz instead of 1 kHz.

For the future, a further modification to CCIR 468-1 is suggested:


The paper includes a schematic and explanation for a transistor-based filter.  7 LP filters are followed by a single HP stage, with a little "feed forward" (R8) to "increase the attenuation in the 9kHz area".

The paper lists about half a dozen test equipment manufacturers who have implemented the new system and goes on to suggest that "CCIR/ARM" is:


Using the Dolby approach, with its average responding meter, seems a curious choice to me for this application, where scratches and dust lead to impulsive noise.  This difficulty was mentioned in an earlier passage.  The bold emphasis is mine.

My summary of the current undesirable situation

The above text highlights the problem I perceive with any noise detection arrangement which does not use a carefully crafted quasi-peak detector, no matter how good its weighting curve is:


In late 2009, it seems this is what we have: some people use A weighting with RMS detection, some use Dolby's so-called CCIR-AMS system and others use the true CCIR-486 system (the standard has been superseded by others with the same technical characteristics).

To me and a number of other people, Dolby's concerns about test equipment costs and numerically larger noise figures are inconsequential and do not at all justify the introduction of this "CCIR-AMS" system to further muddy the waters as the industry should, ideally, to our way of thinking, be abandoning A weighting and unweighted noise figures and adopting CCIR-486 as the sole method of measuring background noise in acoustic settings where human perception is the most important concern.

Tomlinson Holman's "Noise in Audio Systems"

Tomlinson Holman, of Apt Corporation, Cambridge Massechusetts, presented a comprehensive 22 page paper "Noise in Audio Systems". #Holman-1978  This contains lots of discussion about weighting curves and detector systems.

He presents a graph showing the CCIR-468 curve moved down by 6dB, according to the Dolby proposal discussed above.  Here are some quotes:




On detector systems, he writes:


He then discusses loudness measurement according to the work of Eberhard Zwicker.

DIN 45405 was updated on 1983-11-01.  There do not seem to be specific version numbers, but in 2009-09, this 1983 version was still current at DIN.  A 2008 list of standards includes the following:


"Ersatz" means "replace".  Although the standard is in German, its English title is Noise Level Measurements in Sound Systems.  In 2009-09 it is one of 342 standards issued by DIN committee DKE/K 742.

I purchased a copy.  It appears to be a replication of CCIR 468-4, but without any circuit diagrams.  The standard is dated November 1983 and at the top right corner has a space between the numbers: DIN 45 405.  The introductory sentence, in German, and via a rough translation is:


I think this is based on CCIR 468-3, which I think is from 1982.  The original printing of CCIR 468-3, as included as Appendix 1 of IEC 268-1 2nd ed. (1985) contains erroneous figures 44 and 50 in table 2, at the top of the 50 and 100 ms % lower limited values, whereas the correct values are 50 and 58.  The correct values were given by an IEC amendment 1 in 1988, but DIN 45405 of 1983 has the correct values.

Gerhard Steinke's "Important CCIR-Standards for the Audio-World"

Gerhard Steinke, in #Steinke-1984, reports on CCIR 468-3.  His affiliation is with the East German RundFunk- und Fernsehtechnisches Zentralamt der Deutschen Post, Berlin and he is chairman of CCIR working group 10-C, which is concerned with the audio frequency characteristics of sound broadcasting signals.  The measurement of noise voltage is "the most essential part" of 10-C's work.


(I wonder whether distance, language and cultural differences - and cold-war tensions - have contributed to CCIR /  ITU standards being less accepted in North America than they otherwise might have been.  I think "i. a." means inter-alia - "among other things".)

This paper includes the essential text, graphs and diagrams of CCIR 468-3, with some notes of interest, including the following.

Weighting curves

I notice two things here.  Firstly, the tests were done with loudspeakers ("listening room") and secondly the bandwidth of the system was assumed to be limited to 15kHz.

My personal experience is that as an adolescent, I could hear to 17kHz.  Considering the listening audience for much music and other program material in recent years includes adolescents using in-ear headphones, it could be argued that this curve would ideally extend beyond its current sharp cutoff around 15kHz.

Quasi-peak metering

The above, and the following, seems to indicate that Niese developed a noise detector which the CCIR judged to be superior to the quasi-peak detector they chose for 468-3, and that in the future, when digital technologies became more available, that perhaps the Niese detector could be implemented.


He continues to discuss the need for quasi-peak metering:


On the how measurements vary with weighting filter and detector system:


.

IEC 268-1 2nd edition is released, using the same weighting curve ("Q" in Wilms-1970) and the same quasi-peak metering circuit as was then specified in CCIR 468-3.

In 2009-09, this remains current and is known as IEC 60268-1 Ed. 2.0. Sound System Equipment, Part 1, General.  The first few pages are available as a preview. This 2nd edition of 1985 replaces IEC 286-1 (1968) and its three supplements: 268-1A (1970), 268-1B (1972) and 268-1C (1982).

According to #Weeks-1999, IEC 268-1 2nd edition was in effect copied into the British Standard BS 6840-1 (6840: Part 1) in 1993.

In 1986, the most recent (as at 2009-09) version of CCIR 486 was released:


There is no recipe or circuit for constructing the quasi-peak detector, but there is a note that two circuits (I would think diode ->|- capacitor || resistor to ground) circuits could be connected one after the other:


There is an undated article at Lindos Electronics which also mentions a two-stage approach.  This company, under the direction of Peter Skirrow, develops software which implements this quasi-peak algorithm, so the comment is presumably well informed:

ITU-T J.16 Measurement of Weighted Noise in Sound-Programme Circuits is released.  A 1993 version is freely available at  http://eu.sabotage.org/www/ITU/J/J0016e.pdf (and in rough text and Postscript as 3_6_05 at: http://stuff.mit.edu/afs/net/dev/reference/ccitt/1988/ ). 

This consists of a direct copy of CCIR 468-4 of 1986.  It is the same as the version I bought from the ITU, except that this J.16 copy has a little extra information - that the extraordinarily elaborate circuit "constant current realization of the weighting circuit"  was developed here in Australia.

Richard Lee has informed me that during the 1980s (maybe before and after) that in the UK, Germany and I think other European countries, it was common to use the CCIR 468-4 filter, but with a BBC PPM, rather than the quasi-peak meter.  To do: confirm this and get a more exact description.

According to #Weeks-1999 (page 7-33), CCIR 468-4 was released in this year.  Weeks cites it as: CCIR Recommendation 468–4, Measurement of Audiofrequency
Noise Voltage Level in Sound Broadcasting. In: Green Book, Vol. X, Part 1, CCIR, Geneva (1990). This was retitled in 1994.

According to #Weeks-1999 (page 7-33), CCIR 468-4 was retitled: ITU/R Recommendation BS 468–4:1994 

IEC 268-4 2nd edition (microphones) replaces IEC 268-4 1st edition (1972).  The 2nd editions section 16: Equivalent sound pressure level due to inherent noise uses CCIR 468-3 weighting (the "Q" curve in Wilms-1970) and quasi-peak metering. 

It does this by reference to IEC 268-1 (2nd edition, with two amendments in 1988) which contains, as an appendix, all the details from CCIR 468-3.

After this was released, IEC standards had 60000 added to their numbers, so the above became IEC 60268-4 2nd edition. 

The 2nd edition was replaced in 2003 with a 3rd edition, which switched to A weighting with quasi-peak metering.

Stephan Peus' paper "Measurements on Studio Microphones"

Stephan Peus (Neumann) wrote a paper [#Peus-1997] which includes a history of noise measurement standards, with particular reference to using these to determine the equivalent acoustic noise level of microphones.  There are other items of interest regarding microphone testing which I have not quoted.  The quotes below concern noise weighting and detection in general.

Unlike many of the papers cited here, which are only obtainable from the AES E-library, a version of this is available freely on the Neumann website. 

Some quotes from Section 3 on self-noise:






Figure 5 shows CCIR-486-2 "dBq" self noise figures for 12 studio condenser microphones alongside their figures for DIN/IEC 651 "db-A" (A-weighting and RMS detection).  7 of the former are 11 dB higher, 4 are 12 dB higher and one is 9 dB higher.

Vivian Weeks provides a comprehensive description of many audio standards, in #Weeks-1999 in the Audio Engineer's Reference Book 2nd ed.  His "recipe" for constructing a quasi-peak metering system to comply with CCIR 468-2, 468-3 or 468-4 is (page 7-31):

It is not clear to my why one would bother with a separate R-C filter with a time constant of 150 ms except to avoid driving a physical meter movement with short pulses resulting from the rapid (1.6ms) rise time of the first section.  If this description is valid, then taking the long-term (say 2 second or more) average of the output of the second stage would produce identical results to doing the same for the output of the first stage alone. 

So, for generating a single reading from multiple seconds of audio, perhaps we should simply take the average of the first section.

On reading the metering system, Weeks describes a technique which I think is not part of the CCIR 468 or any other standards, and would produce more than a single numeric result:


This raises an interesting question.  If the noise is generally quiet, but on average every minute there is a loud spike of noise, how does one generate an "average" noise figure.

A more thoroughly detailed specification would specify exactly how to create a single figure from an audio signal, such as how many seconds to measure, how to generate the single figure etc.  For instance if ten 10 second measurements produced values with significant variation, with, on average, one such value being significantly worse than the rest, then it would not be realistic to report the average of the dB values as the final result.

Weeks cautions against confusing the Dolby (1978) so-called CCIR-ARM method with the CCIR 468-2/3/4 method.  He also states:


CISPR (The International Special Committee on Radio Interference) is part of the IEC.  According to the IEC list of withdrawn publications, CISPR 6 was published in 1976 and withdrawn in 1977, and is replaced by CISPR 16-1 (1993).  It seems that the most recent (2009-09) such document is CISPR 16-1-1, (1st edition, 2003).

In 2009-09, DIN 45405 (Noise level measurement in sound systems) is available from DIN and is unchanged from  1983-11.

IEC 60268-4 3rd edition (microphones) is released.  Amongst the differences is the requirement for "A" weighting, with quasi-peak metering.  This combination had never been used or suggested by anyone.  According to the AES SC-04-04 meeting notes of 2005-10:



The 3rd edition is current in 2009-09 but a 4th edition is close to being finalized.  A draft of the 4th edition (available to members of AES SC-04-04) shows that weighting will be according to the psophometric curve ("Q" in Wilms-1970), as specified in IEC 60268-1, which includes the text of CCIR 468-3.

IEC 60268-4 Microphones Edition 4.0 is released, replacing the 3rd edition of 2004. 

I have not seen this, but I am reliably informed that in the section called (at least in the 3rd edition) "Equivalent sound pressure level due to inherent noise" that the quasi-peak method is now the required method of measurement.  I am informed that it "now requires 468-4 and has db(A) as optional, rather than the other way round as it had been for a few revisions." 

This is a major step forward in encouraging microphone manufacturers to use the quasi-peak technique when specifying the microphone's inherent self-noise in a manner which properly reflects how it disturbs human listeners.

At some stage, once I have an English translation of #Belger-1954, I will review all this material and attempt to trace these events.

In the following section, the only standards which really matter are CCIR 468-4 and those which cite it, or include its provisions.  These are the "current" standards.  In the "historical" section are other standards which were either unrelated to CCIR 468-4, or which lead to its development.

In the "current" standards section I also discuss the various dBxxx terminologies which may be used to denote particular noise measurement methods.

...

...

I can quote your viewpoint here, or link to it somewhere on the Web.

My viewpoint is on a separate page: vp-rw/ .

Flutter

A related question of trying to measure audio impairment in a way which reflects human subjective experience concerns flutter (rapid pitch variation).  Fortunately this is not a problem with digital systems.  A review of the work in this field is:   Development of a Standard Measurement to Predict Subjective Flutter McKnight, J.  Ampex Corporation, Sunnyvale, CA, USA IEEE Transactions on Audio and Electroacoustics, Mar 1972, Volume: 20,  Issue: 1 pp 75- 78.  The IEEE link is here and the paper is freely available here: http://recordist.com/studer/FlutterReferences.pdf  This paper refers to a 1958 paper by Ernst Belger on flutter measurement:


This is also included in the FlutterReferences.pdf mentioned above.  There's nothing in these papers of direct interest to audio noise measurements, but it does give a full institutional affiliation for Ernst Belger, at least in 1958.

Ernst Belger

I am keen to find out more about Ernst Belger.  The above paper lead me to search for Belger Rundfunktechnik to find out more about this researcher.  I found the following:


This contains a photo, from which I created the above somewhat retouched image, and the following biography:


One paper he co-authored is cited (ref 10) in the BBC's 1967 report on digital audio (Pulse-code modulation for high-quality sound-signal distribution ...":


The full journal name is : "Fernmeldetechnische Zeitschrift" (Telecommunication Technology Magazine).  These can sometimes be found in libraries in Germany, I guess, and original copies can sometimes be found via eBay and http://www.bookfinder.com.

These are papers - this list does not include standards.  Many of these documents are only available at the AES E-library, which costs USD$125 a year to access. 

Steudel-1933 Ulrich-Steudel-1933-graphic-scan.pdf (13 pages.)  Steudel, Ulrich: Über Empfindung und Messung der Lautstärke (On the feeling and measurement of loudness), Hochfrequenztechnik und Elektro-Akustik 41, [1933] booklet 4. S.116.  I listed this title as "Über die Empfindung . . ." but in June 2014 I received a graphic scan of this article, thanks to a researcher finding it in the Library of Congress and now know the title does not contain "die".  #Belger-1953 cites this article extensively and erroneously states that it is from 1937.  

Another citation may be: Steudel. U. 1933 Über die Empfindung und Messung der Lautstärke. Zeitschrift fur Hochfrequenz Technik und Electrotechnik 41: 166 ff.  A Google Books page has a somewhat different citation: Title: Über Empfindung und Messung der Lautstärke; Author: Ulrich Steudel; Publisher: Akad. Verlag-Ges., 1933; Length: 15 pages.  The PDF above matches this later citation, lacking "die" in the title.

An English translation of this important paper was kindly provided in February 2016 by Harvey A. Smith, Professor Emeritus of Mathematics at Arizona State University: https://emerituscollege.asu.edu/sites/default/files/ecdw/EVoice10/mythsandmiracles.html.

As a PDF: Ulrich-Steudel-1933-English-translation-Dr-Harvey-A-Smith.pdf and Word file: Ulrich-Steudel-1933-English-translation-Dr-Harvey-A-Smith.doc .

BBC-1947-12 BBC-1947-12-G.036.pdf  Impulsive Interference in A.M. and F.M. , BBC Research Department Report G.036, Serial 1947/12 AKA Impulsive Interference in AM and FM.  With a 1951 corrigendum.  Authors: D. Morris, G.F. Newell and J.G. Spencer.


BBC-1948-24 BBC-1948-24-G.040.pdf  Experimental Correlation Between Aural and Objective Parameters of Electrical Noise, BBC Research Department Report G.040, Serial 1948/24. 


CCIF-1949 #BBC-1968-EL-17 cites this as: C.C.I.F., Paris, 1949, Vol IV (Yellow Book), pp. 186 - 193.  (I have not seen this - I am quoting the 1968 BBC report.)


Maurice-1950 Maurice, D., Newell, G. F. and Spencer, J. G. 1949 Electrical Noise, Wireless Engr., 1950, XXVII, 316, pp 2 - 12.  (I have not seen this, I am quoting the 1968 BBC report.)


Mangold-1952  H. Mangold, Grundlagen der Geräuschspannungsmessung [Fundamentals of psophometric audio noise measurements], Rohde und Schwartz Mitt., no. 1, pp. 21-36 (1952).  I have not seen this paper.  According to #Wilms-1970, this paper mentions (on page 24) that AT&T proposed the P curve to the CCIF, presumably at or before 1949 when CCIF first adopted it.  Hertz-1977 also cites it, and it is possible that a 1952 chart in this paper comes from Mangold-1952, since it is the only 1952 reference.  #Belger-1953 also cites this paper.


Belger-1953 Ernst. Belger, Über die Messung und Bewertung von Störgeräuschen [On the measurement and weighting of interfering noise], Tech. Hausmitt. NWDR 5, 3/4 pp 51 - 59 (1953). The Wikipedia article for Störgeräusch:


is quite detailed.  Much of the progress in this field has been due to German workers with their tremendous patience, attention to detail and insight into the subtle aspects of physical reality.  This article gives an English translation of Störgeräusch as "interfering noise" but I understand that it also means "disturbing noise".  Therefore, an alternative English translation of the title of Ernst Belger's paper might be "On the measurement and weighting of disturbing noise".

Thanks to Anne Runkel and Dr Hans-Ulrich Wagner  (http://www.hans-bredow-institut.de/en/node/2128) I now have a copy of this paper:  Belger_1953-graphic-scan.pdf .  The pages are graphic images and do not contain text data.  (Still, Google has OCRed the entire document for its search engine database!)

Here is another scan of the article with OCRed text as part of the PDF:  Belger_1953-with-German-text.pdf .

In June 2013, thanks to an anonymous translator, I have prepared an English translation of this paper, which is perhaps the most important in this field.  In July I improved this translation with the help of Hans-Martin Burmeister:


Ernst Belger reviews the history and describes how he and his colleagues devised a suitable weighting curve (filter to accentuate the frequencies according to their ability to disturb) and how they chose a peak detector with specific charge and discharge characteristics, coupled to a slower positive movement and then return towards zero response of a physical meter.  It is my impression that Belger's weighting curve, or something closely based on it, is used in the core standards of this field: CCIR 468 and its derivatives.  I don't fully understand the experimental techniques described in this paper. 

The full journal title is:


which means:


The form of this reference is based on its citation by both the 1968 BBC paper and the 1970 Wilms paper.  Some links regarding where NWDR papers might be found:



Belger-1954 Ernst Belger, Über die Messung und Bewertung von Störwirkung von Geräushen, [Tr: On the measurement and weighting of interference effect of noise] Fernmedetech. Z., 1954, 7, 1, pp. 25 - 32.  The full title of the journal is:


I purchased the twelve 1954 issues and in July 2013, thanks to Hans-Martin Burmeister, we have an English translation.

• Belger-1954-English.pdf (1.5MB).
  
  
• Belger-1954-graphic-scan.pdf (30 MB) A purely graphic scan of the article - no copy and paste of text.
  
  
• Belger-1954-German.pdf (1.4 MB) PDF with the German text (which can be copied and pasted), made from the following Word file.
  
  
• Belger-1954-German.doc (1.6 MB) Word (2003) file of the OCRed German text, with figures.
  
  
• Belger-1954-figures.zip (1.7 MB) 300DPI PNG files of the figures. (I have 600DPI TIFFs if anyone wants them.
  

Niese-1957 (as cited by the 1968 BBC report) Niese, H., Vorschlag für einen Lautstärkemesser zur gehörrichtigen Anzeige. (Translation: Suggestion for a volume meter with weighted loudness reading) Hochfrequenz. Tech. Electroakust., 1958, 66, 4, pp. 125 - 139.


Niese-1958 Niese, H., Vorschlag für einen neuen Schallpegelmesser
Hochfrequenztechnik und Elektroakustik (Proposal for a new sound level meter) 66 (1958), 4, 125.


CCITT-1960 Psophometeter for Broadband Circuits CCITT Recommendation P53, New Delhi (1960)  Red Book, Vol 5, pp 123 - 133. (I have not seen this, I am quoting Wilms' 1970 paper.)


Langmuir-1962  http://www.aes.org/e-lib/browse.cfm?elib=834  A New ASA Standard Sound Level Meter.  Langmuir, D. Bruce, JAES Volume 10 Issue 4 pp. 318-323; October 1962.


WG-8-1966 Proposal for the Extension of the Precision SLM to an Impulse SLM IEC TC 29/WG-8, Prague (1966).  I have not seen this document.  I am quoting from Wilms-1970, reference 14, which he cites on page 653.   ???????????  More info


BBC-1968-EL-17  BBC Engineering Division, Research Report no EL-17 The assessment of noise in  audio-frequency circuits.  http://downloads.bbc.co.uk/rd/pubs/reports/1968-08.pdf .


Reichardt-1968  W. Reichardt and K. Notbohm, Anforderungen an ein Gerät zur Messung des Pegels praktischer impulshaltiger Schalle (Measurement of practical impulsive sound levels), Acustica 20, 159 (1968).  I have not seen this paper. I am quoting reference 10 of Wilms-1970.  I have been informed that the authors were at University of Dresden.


Smith-1970a  http://www.aes.org/e-lib/browse.cfm?elib=1515 Design Considerations of Low-Noise Audio Input Circuitry for a Professional Microphone Mixer.  Smith, A. Douglas; Wittman, Paul H. Shure Brothers, Inc., Evanston, IL JAES Volume 18 Issue 2 pp. 140-156; April 1970.


CCIR-1970-398-1 Measurement of Audio-Frequency Noise in Broadcasting and in Sound Recording Systems CCIR Report 398-1, Study Program 2A/10, New Delhi 1970.  I haven't seen a copy of this.  I am quoting Wilms-1970.


Wilms-1970  http://www.aes.org/e-lib/browse.cfm?elib=1460  Subjective or Psophometric Audio Noise Measurement: A Review of Standards.  Wilms, Herman A, National Institute for Radio- and Filmtechnique, Brussels, Belgium.   JAES Volume 18 Issue 6 pp. 651-656; December 1970.


Smith-1970b  http://www.aes.org/e-lib/browse.cfm?elib=1469 Comments on "Subjective or Psophometric Audio Noise Measurement: A Review of Standards". Smith, A. Douglas, Shure Brothers Inc., Evanston, IL, JAES Volume 18 Issue 6 p. 677; December 1970.


Steffen-1972 Steffen, E.; Untersuchungen zur Geräuschspannungsmessung (Investigations for noise voltage measurements) Techn. _tt. d. RFZ 16 (1972) 3, 81-84.  A more correct citation is:


Hertz-1973 http://www.aes.org/e-lib/browse.cfm?elib=1728 A New Solid State Peak Programme Meter, Hertz, Bent F., NTP Audio, Copenhagen, Denmark AES Convention:44 March 1973.


CCIR-1974-398-2 OCIR-Report 398-2, Measurement of Audio-Frequency Noise in Broadcasting and in Sound Recording Systems. CCIR, XIIIth Plenary Assembly, Geneva 1974, Vol. X.  I haven't seen this report.  I am quoting #Steinke-1984 .


Hertz-1977 http://www.aes.org/e-lib/browse.cfm?elib=3161 Psophometric noise measurement on audio equipment, Bent Hertz, Danish Broadcasting Copenhagen. (Note, the paper itself is titled Noise measurement on audio equipment.) 


Dolby-1978 http://www.aes.org/e-lib/browse.cfm?elib=3218 CCIR-ARM: A Practical Noise-Measurement Method.  Dolby, Ray; Robinson, David; Gundry, Kenneth J., Dolby Laboratories, Inc., San Francisco, JAES Volume 27 Issue 3 pp. 149-157; March 1979. (The use of "CCIR" in the name of this proposal is not authorised by CCIR or its its successor, the ITU-R.)


Holman-1978  http://www.aes.org/e-lib/browse.cfm?elib=2926 Noise in Audio Systems Tomlinson Holman, Apt Corporation, Cambridge Massachusetts, AES Convention November 1978.


Jakubowski-1980 Jakubowski, H.; Quantisierungsverzerrungen in digital
arbeitenden Tonsignal_bertragungs- und verarbeitungsstemen (Distortions in digital sound systems for transmission and processing) Rundfunktechnische Mitteilungen 24 (1980) 2, 91-92


Steinke-1984  http://www.aes.org/e-lib/browse.cfm?elib=11778  Important CCIR-Standards for the Audio-World, Steinke, Gerhard, RundFunk- und Fernsehtechnisches Zentralamt der Deutschen Post, Berlin, German Democratic Republic (GDR).  AES Convention: March 1983.


Peus-1997  http://www.aes.org/e-lib/browse.cfm?elib=7162  Freely available at: http://neumann.com/download.php?download=lect0020.PDF Measurements on Studio Microphones,  Stephan Peus, Georg Neumann GmbH, Berlin, Germany. AES Convention:103 (September 1997).


Skirrow-1999 Audio measurements and test equipment, Peter Skirrow (home page) of Lindos Developments, section of Audio Engineer’s Reference Book, 2nd ed. Michael Talbot-Smith, editor. Focal Press, 1999. Pages 3-94 to 3-108.


Weeks-1999 International standards for sound systems and equipment, Vivian Weeks, formerly of BBC Design and Equipment Department, section of Audio Engineer’s Reference Book, 2nd ed. Michael Talbot-Smith, editor. Focal Press, 1999. Pages 7-19 to 7-33.


Bohn-2000 Bewildering Wilderness - Navigating the complicated and frustrating world of audio standards,  Dennis A. Bohn, President of Research and Development, Rane Corporation.  Sound and Video Contractor, September 2000, pages 56 to 64.  http://www.rane.com/pdf/bewilder.pdf .

http://www.aes.org/e-lib/browse.cfm?elib=1469

2009-09-07:  Initial version, password protected - only a few people saw it. 

2012-03-01:  A few updates and made it public.

2013-06-18:  Added English translation of Ernst Belger's 1953 paper.  Noted that IEC 60268-4 has 468-4 as the primary noise measurement technique.  Corrected several items thanks to help from a fellow in Germany, including the correct name for the journal in which Ernst Belger's 1954 paper appeared.  Added this Update history section.  I did not attempt to revise the main body of material.  In days following this I added some PDFs from the BBC and the ITR-R and made a number of minor improvements.

2013-07-02:  Added graphic scan and German OCR version of Ernst Belger's 1954 paper.  Improved the translation in his 1953 paper.

2013-07-11: Added an English translation of Ernst Belger's 1954 paper.

2014-06-21: Added a graphic scan of Ulrich Steudel's 1933 paper, a translation of one sentence from page 123 and the two waveform files and a graphic depicting them.  Corrected some typos and replaced some rough translations (from Google) with real translations from Hans-Martin Burmeister.

2016-02-12: Added an English translation of Ulrich Steudel's 1933 paper.