Errata in ISO 14257
These are the 4 entries of the errata registry on ISO 14257:2001, in the order the registry lists them. The registry’s introduction says how each claim is checked against the printed page and what the status of an entry means.
ISO 14257:2001, Annex C (an example that corrects one of its two results)
Section titled “ISO 14257:2001, Annex C (an example that corrects one of its two results)”- Location: C.1 on printed page 17 (PDF page 27), against Tables C.7 and C.9 on printed page 23 (PDF page 33) and Tables C.8 and C.10 on printed pages 23 and 24 (PDF pages 33 and 34).
- The print: C.1 lists the four things that hold in the example, the second of which is that “the experimental reference curve of the sound source is known and used for correcting the values measured in the workroom”. Table C.5 is then headed “Values of , in octave bands (corrected for background noise)” and Table C.6 “Values of … corrected for background noise and using the experimental reference curves of the source”.
- The problem: the four result tables are not all computed from the same curve. Every value of in Tables C.7 and C.8 follows from the corrected curve of Table C.6, and every value of in Tables C.9 and C.10 follows from the uncorrected curve of Table C.5. Swap either one for the other and 28 of the 36 printed results leave the rounding of the table they are printed in.
- Evidence: the two curves are printed in full, so both routes can be run. Taking Table C.6 as printed, the middle range at 1 kHz gives = 4.39 dB against the printed 4.4 and = 6.73 dB against the printed 7.3; taking Table C.5 instead gives 4.73 dB and 7.29 dB. The same split holds in all three distance ranges and all six octave bands, and in the two A-weighted pink-noise tables. Verified on PDF pages 27, 31 and 33 (printed pp. 17, 21 and 23) of EN ISO 14257:2001 as published in BS EN ISO 14257:2001.
- Consequence for the standard’s own tables: none for , which is a slope and is barely moved by a correction that is nearly constant with distance. For , which is a level, the difference reaches 1.4 dB in the near range.
- Library behaviour:
corrected_distribution_valueapplies Annex B when it is asked to and never on its own, so the caller chooses which curve each descriptor is taken from. The conformance row “ISO 14257:2001 Annex C (C.1 against Tables C.7 and C.9)” pins the split, andtest_the_annex_corrects_the_decay_but_not_the_excessintests/room/test_spatial_decay.pyholds both numbers so that neither can drift. - Status: not reported.
ISO 14257:2001, Equation (5) against Equation (8) (a rounded logarithm)
Section titled “ISO 14257:2001, Equation (5) against Equation (8) (a rounded logarithm)”- Location: Equation (5) of 6.3 on printed page 9 (PDF page 19) and Equation (8) of 6.4.3 on printed page 10 (PDF page 20).
- The print: Equation (5) opens with the factor in front of the least-squares slope; Equation (8), one page later, divides by .
- The problem: the two are the same conversion, from a rate per decade to a rate per distance doubling, written twice with different precision. is 0.301 03, so the printed 0.3 is 0.34 % small, and a document that prints the exact form on one page has no reason to round it on the previous one.
- Evidence: the two equations on facing pages, both reproduced in the entry above from the printed tables. Verified on PDF pages 19 and 20 (printed pp. 9 and 10) of EN ISO 14257:2001.
- Consequence for the standard’s own tables: none that the printed rounding can show: the worked example of Annex C is tabulated to one decimal and 0.34 % of a 4 dB slope is 0.014 dB.
- Library behaviour:
DECADE_TO_DOUBLINGcarries the printed 0.3, because the printed constant is what reproduces the printed results, andlevel_excess_atdivides by where Equation (8) prints it. Do not unify them. - Status: not reported.
ISO 14257:2001, Equation (4) against Table 1 (a second rounded constant)
Section titled “ISO 14257:2001, Equation (4) against Table 1 (a second rounded constant)”- Location: Equation (4) of 4.2.3 and Table 1 immediately beneath it, both on printed page 4 (PDF page 14), against the last column of Table C.6 on printed page 21 (PDF page 31) and Table C.10 on printed page 24 (PDF page 34).
- The print: Equation (4) closes dB dB, and Table 1 gives the of the A-weighted pink-noise reference spectrum as ; ; ; ; ; dB at 125 Hz to 4 kHz.
- The problem: the 6.2 dB is the energy sum of the , which is what normalizes the weighted spectrum back to unit total so that a flat curve comes back unchanged, and the six printed sum to 6.251 5 dB, which rounds to 6.3 and not to 6.2. The printed constant is 0.051 dB short of the printed table. The two are roundings of the same curve made separately: the are the A-weighting of IEC 61672-1 at the six octave centres, ; ; ; ; ; dB, printed to one decimal as that standard tabulates them, and the energy sum of the unrounded curve is 6.23 dB, which prints as the 6.2 of Equation (4). Each rounding is right on its own and the printed equation is not: evaluated as printed, it returns a flat curve 0.05 dB high. It is the same kind of slip as the rounded of the entry above, a constant printed to one decimal where the document computes to more, and Equation (3), of which Equation (4) is the special case for the Table 1 spectrum, carries the same normalization exactly, as the logarithm of its denominator.
- Evidence: Equation (4), its key and Table 1 read on the page in four printings, which agree character for character: ISO 14257:2001(E), PDF page 10 (printed p. 4); EN ISO 14257:2001 as published in BS EN ISO 14257:2001, PDF page 14 (printed p. 4); UNE-EN ISO 14257:2002, PDF page 9 (printed p. 9); and DIN EN ISO 14257:2011-11, PDF page 13 (printed p. 9), whose national foreword lists the technical errors corrected in the German text and does not name this one. The annex settles which constant it was computed with. Run from the printed Tables C.2 to C.4 through Annex B without rounding, the last column of Table C.6 comes back with the six printed and their own sum, 6.251 5 dB, inside the printed rounding at all 11 positions (worst 0.041 dB, departures of both signs), and with the printed 6.2 dB one unit high in the last place at 6 of the 11 (worst 0.092 dB, every departure positive, from +0.011 to +0.092 dB). Table C.10 does the same: +0.040, +0.004 and +0.045 dB with the sum, +0.091, +0.055 and +0.097 dB with 6.2. The unrounded A-weighting with its own sum lands the same 14 cells inside the rounding (worst 0.047 dB), so the annex does not say which of the two exact normalizations it used, only that it used one; with the printed 6.2 dB neither weighting does (8 and 9 of 14 outside). Over the printed, already rounded octave columns of Table C.6 the split is 3 of 11 against 7 of 11. Table C.8, a slope, does not see the constant, and the A-weighted 115.7 dB of Table C.2 adds it back. Verified on PDF page 14 (printed p. 4) of EN ISO 14257:2001 for the equation and the table, and on PDF pages 31 and 34 (printed pp. 21 and 24) of the same document for the two annex tables.
- Consequence for the standard’s own tables: none for the annex, which was normalized exactly. What carries the 0.051 dB is every evaluation of the printed equation, which stands that much above the annex on each frequency-normalized value and can move a printed cell of Tables C.6, C.10 and C.12 by one in the last place, always upward, and never by more.
- Library behaviour:
NORMALIZED_OFFSET_DBcarries the printed 6.2, on the same rule as the entry above and as the 11 of Equation (2): the printed constant is what a reader checking against the page will use, and four printings print it. The exact normalization is Equation (3) with the Table 1 weights as the machine spectrum, whichspectrum_distribution_valuecomputes, so the annex’s reading is available without a second constant. The conformance row “ISO 14257:2001 Annex C, Table C.6 last column” judges the printed constant against the annex at 0.1 dB and says why, the row “ISO 14257:2001 Eq. (4) against Annex C, Table C.6 last column and Table C.10” records that the exact sum lands all fourteen values inside the printed rounding and the printed constant does not, andtest_the_annex_normalized_with_the_table_one_sum_and_not_the_printed_offsetintests/room/test_spatial_decay.pyholds both counts so that neither the constant nor the tolerances can drift. - Status: not reported.
ISO 14257:2001, Annex C, Tables C.11 and C.12 (results Equation (8) does not give)
Section titled “ISO 14257:2001, Annex C, Tables C.11 and C.12 (results Equation (8) does not give)”- Location: Tables C.11 and C.12 on printed page 24 (PDF page 34), against Equation (8) of 6.4.3 on printed page 10 (PDF page 20) and Table C.5 on printed page 21 (PDF page 31).
- The print: Table C.11 gives in six octave bands at 4 m, 10 m and 30 m from the source: 6,2; 4,0; 4,4; 5,2; 5,3; 3,9 at 4 m; 7,8; 5,4; 6,4; 6,9; 7,7; 5,8 at 10 m; and 10,6; 7,4; 9,3; 7,2; 9,2; 6,1 at 30 m. Table C.12 gives 4,8; 6,8; 8,0 dB for the A-weighted pink-noise spectrum at the same three distances.
- The problem: Equation (8) is the only thing in the document that defines , and it does not produce these numbers. Evaluated over the three distance ranges the same annex uses, 2 m to 5 m, 5 m to 24 m and 24 m to 48 m, on the printed of Table C.5, it departs from Table C.11 by up to 1.55 dB, and the printed rows are not an Equation (8) result for any decay rate: solving the equation for the the middle row would need gives values from -28.9 dB to +20.7 dB per distance doubling. What the three rows are is a reading of the printed measurement at a microphone position, through Equation (6) rather than Equation (8): the 4 m row is Equation (6) at the 4 m position of Table C.5, in all six bands; the 10 m row is the mean of Equation (6) at the 8 m and 12 m positions that bracket it, in all six bands; and the 30 m row is Equation (6) at the 32 m position with entered as 30 dB rather than 30.103 dB, in five bands of six.
- Evidence: Tables C.10, C.11 and C.12 read on PDF page 34 (printed p. 24), Tables C.5 and C.6 on PDF page 31 (printed p. 21) and Equations (6), (7) and (8) with their keys on PDF page 20 (printed p. 10), all of EN ISO 14257:2001. Seventeen of the eighteen cells of Table C.11 come back from the printed of Table C.5 by the reading above, to the tenth the table prints; the exception is the 125 Hz cell of the 30 m row, which prints 10.6 where that reading gives 10.7. The same digits are printed in the Spanish adoption, UNE-EN ISO 14257:2002, Tabla C.11 on printed p. 30, so this is the annex and not one printing of it.
- Consequence for the standard’s own tables: Tables C.11 and C.12 only. The decay rates and excesses of Tables C.7 to C.10 are unaffected, and nothing downstream computes with C.11.
- Library behaviour:
level_excess_atimplements Equation (8) as printed and returns the values above rather than the table’s. The conformance row “ISO 14257:2001 Annex C, Tables C.11 and C.12” records the departure so the table is not mistaken for an oracle of Equation (8). - Status: not reported.
Created and maintained by José Manuel Requena PlensORCIDAbout the maintainerRelease notesChangelogAll projects