• Octaves and third-octaves

    Octaves, thirds of octaves and decades are divisions of the spectrum used in audio, since they work in the same way as the ear, increasing their width in Hertz as frequency increases, and decreasing it when frequency decreases. They are used for the definition of the width of filters, as well as for the spectral analysis.

    What is an octave?

    An octave is a frequency band that lies between one frequency and another that is twice the frequency, such as 1000 to 2000 Hz, or 707 to 1414 Hz (the latter corresponding to the octave band centered at 1000 Hz1). As mentioned at the beginning, like third-octaves and decades, bandwidth of octave bands doubles when we double the frequency. Thus, the 1000 to 2000 Hz octave has 1000 Hz width (2000 minus 1000), and the 2000 to 4000 Hz octave has 2000 Hz width (4000 minus 2000), while the 4000 to 8000 Hz octave has 4000 Hz width (8000 minus 4000).

    An octave is also the interval between a musical note and the upper or lower octave of the scale, encompassing 12 semitones. Para ayudar a dar una idea de lo que supone una octava a los que no son músicos, es el intervalo entre las dos primeras notas del estribillo de la canción 'Somewhere Over the Rainbow' (la palabra 'somewhere'). Para los no músicos, las notas musicales se repiten cada octava. Por ejemplo, a 440 Hz está el La (A, en notación anglosajona) central estándar, pero a 220 y 880 Hz las notas son también La. Cuando tocamos dos notas con el mismo nombre en un piano espaciadas una octava (por ejemplo, el La con frecuencia fundamental de 220 Hz y el de 440 Hz), la sensación es de reforzar esa nota, mientras que con cualquier otra combinación de notas (por ejemplo un La y un Do), resulta evidente que hay una cierta disonancia.

    Assuming that the audible spectrum ranges from 20 to 20,000 Hz, it could be divided into approximately ten octaves.

    What is one third of an octave?

    One third of an octave (third-octave) is a frequency band that corresponds to one third of an octave (and therefore, three adjacent bands of one third of an octave correspond to one octave band).

    And why a third and not a half or a quarter? The answer to this is to be found in the spectral behaviour of the ear, which we could approximate to a bank of third octave filters.

    Musically, on the equal temperament Western scale, one third of an octave is a 'third' interval, comprising four semitones (or almost, depending on the musical pitch and the way in which the frequency of the third of an octave is calculated 1).

    In spectrum analysis applications, fractional octave values such as thirds are often used, but also others such as half (1/2), 1/4, 1/8, 1/12, 1/16, 1/24 or 1/48 octave, either for a low resolution 'RTA' type analysis, or for the smoothing of a full resolution frequency response. Assuming that the audible spectrum ranges from 20 to 20,000 Hz, it could be divided into approximately 30 thirds of an octave (one-third octave equalizers such as the one shown in the photograph usually have 31 because their effect extends half a third-octave above and below the nominal audible band).

    As a curiosity, scaling the nominal frequencies of one-third octave can help calculate decibel differences without a calculator. For example, for power, we could remember 1250 and 1600 Hz and use them (divided by 1000) as multipliers (or dividers) to perform, an approximate conversion to relative decibels, namely add (or subtract) 1 or 2 decibels. The exact ratios are actually 1.259 and 1.589, respectively, but those nominal frequencies are close enough (specially considering that the ear has trouble distinguishing even 1 dB changes between levels). Similarly, these same frequencies can be used to calculate decibels of number of loudspeakers, voltage or pressure (or even distance to the loudspeaker), although in this case the result will be to add (or subtract) 2 or 4 dB. Naturally, the factor of 2 from 2000 Hz leads to 3 or 6 decibels; likewise, the factor 1 from 1000 Hz leads to 0 decibels.

    And what about decades?

    A decade is a frequency band that lies between a first frequency and another that is ten times greater than this, such as 1000 to 10000 Hz.

    Musically, on the western equal-temperament scale, this assumes an interval of 3 octaves and four semitones and, therefore, 10 thirds of an octave (or almost, depending on the musical tuning and how the frequency of the third of an octave is calculated 1). Assuming that the audible spectrum ranges from 20 to 20,000 Hz, it could be divided into three decades: 20-200 Hz, 200-2,000 Hz (which can be seen in the graph above) and 2,000-20,000 Hz (which, depending on the criteria used, could be labelled as 'low', 'medium' and 'high').

    The width of the different bandwidths can be seen in the graph above: decade, octave and third of octave. As a reference, we have added the keyboard of a standard 88-key piano. The green key corresponds to the central A (La, in italian notation) (of octave number four), with a frequency of 440 Hz, and the grey key corresponds to the central C (Do).

    Center frequencies for octave and third-octave filters

    The centre (midband) frequencies of the octave and octave fractions (third and half) filters are defined in ISO and ANSI standards, as well as and their 'nominal' (rounded) numerical representations.

    The table below shows, for the ten commonly used audio octave bands, the mathematical and nominal (rounded) centre frequencies for 1-octave width filters, in addition to the band limiting (bandedge) frequencies:

    In the case of centre frequencies for one-third octave wide filters, the above standards apply the so called 'preferred values' developed by Charles Renard at the end of the 19th century and which are also used to establish capacitor and resistor value spacing. As with one-octave filters, the centre frequencies and the 'nominal' (rounded) frequencies of the filters are standardized.

    The table below shows the center frequencies for the 31 third-octave bands commonly used in audio, for filters with one-third octave width, in addition to the bandedge frequencies):

    1 The centre frequencies of the octave and octave fraction filters are logarithmically spaced (i.e., they are further and further apart as we move up in frequency) and can be calculated in two standardized ways. The first is in base 10 and the second is in base 2, each with its advantages and disadvantages. We could say that the first form is more technical and the second is more musical. For the text of the article and the tables we have used base 2, the most common. As a reference, the central frequencies calculated in base 10 for octave band filters are: 31.62, 63.10, 125.9, 251.2, 501.2, 1000, 1995, 3981, 7943 and 15849 Hz. It can be observed that octave spacings do not correspond exactly to double or half. The octave third filters are: 25.12, 31.62, 39.81, 50.12, 63.10, 79.43, 100.0, 125.9, 158.5, 199.5, 251.2, 316.2, 398.1, 501.2, 631.0, 794.3, 1000, 1259, 1585, 1995, 2512, 3162, 3981, 5012, 6310, 7943, 10000, 12589, 15849 and 19953 Hz.

    2 Mathematically calculated frequencies have been rounded for ease of reading.

    References: ANSI S1.11: Specification for Octave-Band and Fractional-Octave-Band Analog and Digital Filters
  • Ad

  • News

    inMusic acquires ArKaos

    Fort Lauderdale, FL USA (October 27, 2020) inMusic, the has announced the acquisition of ArKaos.

    Founded in 1996, ArKaos specializes in video processing technologies,... Read more

    Meyer Sound launches Spacemap Go, a tool for spatial sound design and mixing

    Meyer Sound has officially released Spacemap® Go for spatial sound design and mixing. Available now as a free app for Apple iPad (any iPad capable of running the latest iPadOS), Spacemap... Read more

    Bose unveils L1 Pro family

    These three new portable Bose PA systems are designed to give artists choices, suit different styles and audiences, and provide an solution for creators reintroducing live music and sound experiences into venues and online platforms across... Read more

    ClearOne BMA 360 beamforming microphone array ceiling tile

    ClearOne (NASDAQ:CLRO), provider of audio and visual communication solutions, has announced its new BMA 360, reportedly the world’s most technologically advanced Beamforming Microphone Array Ceiling Tile.

    The ClearOne BMA 360 is the world’s first truly wideband,... Read more
  • Recent articles

    Power amplifier modes : stereo, parallel and bridge mono

    In general, two-channel power amplifiers for professional use default to stereo mode. That is, each amplifier channel receives a signal from its input connector and its volume is controlled by... Read more

    Basic electricity formulas

    Although it not specific to sound, we include this document with some basic electricity formulas. They can be found in any electricity textbook, but we have added them to the DoPA Library for reference.

    Ohm's law

    The most basic formulas derive from Ohm's law, which specifies that the electric current between two points is proportional to the potential difference (voltage) between them and inversely proportional to the resistance between them. The formula is:
    I = ———

    where I is the current (intensity) in amps and V is the voltage in volts. Since we use alternating current in audio, we have replaced resistance with impedance (Z, and this could also be resistance R), measured in ohms. Clearing Z and V we have these other two formulas:

    Z = ———
    ... Read more

    Y-cables, looping audio signals through

    This article will explain loop-though connections and "y-cables" for analog audio signals.

    To obtain one or more copies of a signal (for example, to distribute the signal from a mixer to various self-powered speakers or power amplifiers) we use parallel connections. To do this we simply connect each terminal (pin) of the connectors in parallel. That is, 1 to 1, 2 to 2 and 3 to 3 (or tip to tip, ring to ring and sleeve to sleeve in a TRS connector). When we split a signal in two in this way, we refer to a "Y-cable" or "Y" connection, since the division of a signal in two looks like letter "y". Contrary to what it may seem, a y-cable is not a technically incorrect solution, but a correct way of splitting the audio signal. In fact, when a self-powered loudspeaker system or one channel of... Read more
  • PAcalculate app