emission.sound_power_hard_walled

Sound power and sound energy levels of a small movable source in a hard-walled test room, by comparison with a reference sound source: ISO 3743-1:2010 (engineering grade 2).

The room is an ordinary one: nearly empty, with smooth hard walls, no surface anywhere absorbing more than 0,20 of the incident power (4.3), at least 40 m³ and forty times the reference box (4.2). Such a room is not diffuse enough to be read on its own the way ISO 3741 reads a qualified reverberation room, so the room is not modelled at all: a calibrated reference sound source (RSS) stands where the source under test (ST) stood and the same microphones listen to both. The room then drops out of the difference, and the octave-band sound power level of the source is the calibrated power of the reference carried across by the difference of the two mean levels, each corrected for background noise (8.1.4):

The mean levels are energy averages over the $N_{\mathrm{M}}$ microphone positions or traverses (Eq. 10, 11, 12), and when the preliminary survey of 7.4 asked for more than one source location the levels at each position are first energy-averaged over the $N_{\mathrm{S}}$ locations (Eq. 9).

Unlike ISO 3741 and ISO 3747, the background correction is taken once per band from the two means, not position by position (8.1.3):

with three rules around it: above 15 dB there is nothing to correct, from 6 dB to 15 dB Eq. (13) applies, and below 6 dB the correction is fixed at 1,3 dB, “the value for $\Delta L_p$ = 6 dB”, and the band becomes an upper bound that the report has to flag. The same rule gives $K_{1(\mathrm{RSS})}$ from the reference source’s mean, but a band where it is the reference source’s margin that falls short is no upper bound: $K_{1(\mathrm{RSS})}$ enters Eq. (14) with a plus sign, so the capped correction pulls $L_W$ down, not up. 8.1.3 gives the upper-bound reading to the margin of the source under test alone, and such a band only fails the background requirement of 4.5, which is all the library says of it.

A source that emits bursts has a sound energy level instead (8.2): the single event levels of each position are reduced to the level of one event (Eq. 15 or Eq. 16), averaged over the source locations (Eq. 17) and the positions (Eq. 18), corrected by the same rule (Eq. 19), and

Eq. (19) subtracts a time-averaged background level from a time-integrated event level, exactly as ISO 3744:2010 Eq. (21) does, and asks only that both be measured over one integration time $T$. The library reads the background as its exposure over that $T$, $L_{p(\mathrm{B})}+10{\log }_{10}(T/T_0)$, which is the reading under which the insistence on one $T$ does any work (see docs/ERRATA.md), so integration_time_s is required with a background.

Annex A carries either level to the reference meteorological conditions of 101,325 kPa and 23,0 °C with the radiation-impedance correction $C_2=-10{\log }_{10}(p_{\mathrm{s}}/p_{\mathrm{s},0})+15{\log }_{10}[(273,15+\theta )/{\theta }_1]$, ${\theta }_1$ = 296 K, the expression ISO 3741:2010 and ISO 3747:2010 print digit for digit; the static pressure follows from the altitude by Eq. (A.2), which is ISO 3747:2010 Eq. (C.2) (static_pressure_from_altitude). The correction is required above 500 m (8.1.4). Annex B forms the A-weighted totals from the octave bands with the Table B.1 corrections, which are ISO 3744:2010 Table E.2 digit for digit, 63 Hz row included under its footnote.

Clause 9 estimates the uncertainty as ${\sigma }_{\mathrm{tot}}=\sqrt{{\sigma }_{R0}^2+{\sigma }_{\mathrm{omc}}^2}$ (Eq. 22) and $U=k{\sigma }_{\mathrm{tot}}$ (Eq. 23), with $k$ = 2 (9.5), and Table 3 gives the typical upper bound of ${\sigma }_{R0}$ per octave band: 3,0 dB at 125 Hz, 2,0 dB at 250 Hz, 1,5 dB from 500 Hz to 4 kHz, 2,5 dB at 8 kHz and 1,5 dB for the A-weighted level of a flat spectrum. The middle row of the printed table reads “400 to 5 000” under an octave heading, which is a one-third-octave range; the library reads it as the four octaves it spans (see docs/ERRATA.md).

The room itself is qualified in three steps, which check_hard_walled_room evaluates together: its volume against the reference box (4.2), the absorption of its surfaces (4.3), and the acoustic adequacy of 4.4, eight mean octave levels of a highly directional source turned through four horizontal and four vertical orientations, whose largest spread in each band from 125 Hz to 8 kHz may not exceed the Table 3 value. The preliminary survey of 7.4, six microphone positions whose standard deviation decides how many source locations the determination needs (Table 2), is hard_walled_source_locations.

Auto-generated from the source docstrings by scripts/generate_api_docs.py (make api-docs). Do not edit by hand.

check_hard_walled_room(
    orientation_levels: ArrayLike,
    frequencies: ArrayLike,
    *,
    volume_m3: float,
    reference_box_m: ArrayLike,
    absorption_coefficients: ArrayLike | None = None,
) -> HardWalledRoomCheck

Is this room a hard-walled test room? ISO 3743-1:2010, 4.2 to 4.4.

The acoustic criterion of 4.4 turns a highly directional broadband source, directivity index at least 5 dB above 500 Hz, through four horizontal and four vertical orientations, each time taking the mean background-corrected octave level over the microphone positions. The room suits the method if in every band from 125 Hz to 8 kHz the largest difference between any two of the eight means stays within the standard deviation of reproducibility of Table 3. The NOTE allows a source of the type to be tested in place of the directional one, which then qualifies the room for that type only.

4.2 adds the size: at least 40 m³ and forty times the reference box, whose largest dimension is at most 1,0 m in a room of up to 100 m³ and 2,0 m in a larger one. 4.3 adds the surfaces: no portion of any boundary absorbing more than 0,20 at any frequency of interest; Table 1 describes the rooms that meet it in words, and the coefficients are optional here.

Parameters

Returns: The verdict, as a HardWalledRoomCheck.

Raises

hard_walled_source_locations(
    levels: ArrayLike,
    frequencies: ArrayLike,
) -> SourceLocationPlan

The number of source locations from the preliminary survey of ISO 3743-1:2010, 7.4 (Table 2).

For a source with audible discrete tones or narrow bands of noise, at least six fixed microphone positions record the uncorrected level of the source under test, and their standard deviation per band

about their arithmetic mean (Eq. 8) decides the number of source locations: one up to 2,5 dB, two in the same room up to 4,0 dB, and above that two in the same room plus two more in another room of different dimensions that still complies with 4.4.

Parameters

Returns: The SourceLocationPlan.

Raises

HardWalledRoomCheck(
    frequencies: np.ndarray,
    level_range_db: np.ndarray,
    orientations: int,
    volume_m3: float,
    reference_box_volume_m3: float,
    largest_box_dimension_m: float,
    max_absorption_coefficient: float,
)

Qualification of a hard-walled test room (ISO 3743-1:2010, 4.2 to 4.4).

level_range_db is, per octave band, the largest difference between the mean levels of any two of the directional source’s orientations (4.4), and limit_db the Table 3 value it may not exceed. volume_m3 is the room, reference_box_volume_m3 and largest_box_dimension_m the reference box of 4.1, and minimum_volume_m3 / box_dimension_limit_m what 4.2 asks of them. max_absorption_coefficient is the largest sound absorption coefficient of any portion of the boundary surfaces, NaN when none was supplied, against the 0,20 of 4.3. minimum_microphone_distance_m is the $d_{\min }=0,3V^{1/3}$ of 7.3 that keeps the microphones in the reverberant field, stated for the setup rather than judged. The limits are the standard’s and are read from the bands, the room and the box, so a check cannot be built against other ones.

property

The 4.4 criterion: every band within Table 3.

property

Per band, whether the spread stays within Table 3 (4.4).

property

The largest reference-box dimension 4.2 allows in this room.

Returns: 1,0 m in a room of up to 100 m³, 2,0 m in a larger one, in metres.

property

The size criterion of 4.2 on the largest reference-box dimension.

property

The Table 3 standard deviation of reproducibility of each band, in dB.

Returns: The spread 4.4 allows each band.

property

The $d_{\min }=0,3V^{1/3}$ of 7.3, in metres.

Returns: The least distance from the source that keeps a microphone in the reverberant field.

property

The volume 4.2 asks of the room: 40 m³ and forty reference boxes.

Returns: The larger of the two, in cubic metres.

property

Whether the room qualifies on every criterion that was evaluated.

HardWalledRoomCheck.plot(
    ax: Axes | None = None,
    *,
    language: str = 'en',
    **kwargs: Any,
) -> Axes

Plot the orientation spread per band against the Table 3 limit.

Requires matplotlib (pip install phonometry[plot]).

Parameters

Returns: The axes.

property

The 4.3 criterion, None when no absorption was supplied.

property

The volume criterion of 4.2: at least 40 m³ and 40 reference boxes.

Both volumes are rounded to nine decimal places before they are compared: forty times a box measured in decimal metres can come out a last bit above a room volume that equals it, and 4.2 asks for “at least”.

HardWalledSoundPowerResult(
    frequencies: np.ndarray,
    sound_power_level: np.ndarray,
    sound_energy_level: np.ndarray,
    mean_source_level: np.ndarray,
    mean_reference_level: np.ndarray,
    mean_background_level: np.ndarray,
    reference_power_level: np.ndarray,
    background_correction: np.ndarray,
    background_correction_ref: np.ndarray,
    background_margin_db: np.ndarray,
    background_margin_ref_db: np.ndarray,
    c2: float,
    sigma_r0: np.ndarray,
    sigma_r0_a: float,
    sigma_omc: float,
    coverage_factor: float,
    sound_power_level_a: float,
    sound_energy_level_a: float,
    quantity: str,
    microphone_positions: int,
    source_positions: int,
)

Result of an ISO 3743-1:2010 determination in a hard-walled test room.

quantity says which of the two determinations this is: 'power' carries the octave-band sound power level LW (Eq. 14) in sound_power_level with sound_energy_level all NaN, and 'energy' the sound energy level LJ (Eq. 20) in sound_energy_level with sound_power_level all NaN. Both are at the meteorological conditions of the test; the ..._ref properties add the Annex A correction c2, which 8.1.4 requires above 500 m.

mean_source_level is the uncorrected mean level of the source under test, $\overline{L_{p(\mathrm{ST})}^{\prime }}$ (Eq. 10) or $\overline{L_{E(\mathrm{ST})}^{\prime }}$ (Eq. 18), and mean_reference_level that of the reference sound source, $\overline{L_{p(\mathrm{RSS})}^{\prime }}$ (Eq. 11); mean_background_level is $\overline{L_{p(\mathrm{B})}}$ (Eq. 12), NaN where no background was measured, and reference_power_level the calibrated $L_{W(\mathrm{RSS})}$. background_correction and background_correction_ref are $K_1$ and $K_{1(\mathrm{RSS})}$ per band, and background_margin_db and background_margin_ref_db the two margins over the background they were read from (NaN where no background was measured). background_requirement_met and upper_bound are read from those margins and the 6 dB of 4.5, so they are not fields: the first is True only where both margins reached it, and the second marks the bands that 8.1.3 calls upper bounds, where the margin of the source under test was measured and fell below 6 dB while the reference source’s margin met it, so the capped $K_1$ leaves the level too high. A band where the reference source’s margin falls short is not one: the capped $K_{1(\mathrm{RSS})}$ pulls the level down, so it is flagged by background_requirement_met alone, and so is every band when no background was measured.

sigma_r0 is the Table 3 value per band (NaN at 63 Hz, which the table does not reach) and sigma_r0_a its A-weighted row; with sigma_omc the properties give sigma_tot (Eq. 22) and the expanded uncertainty (Eq. 23) for coverage_factor, all NaN without it. microphone_positions is $N_{\mathrm{M}}$ (1 for a traverse), source_positions is $N_{\mathrm{S}}$, and sound_power_level_a / sound_energy_level_a are the Annex B totals of the level that was determined (NaN for the other).

property

Per band, whether both margins over the background reached the 6 dB of 4.5.

False where no background was measured. Each margin is judged settled, as $K_1$ is read from it.

property

U = k sigma_tot per band (Eq. 23), in decibels.

property

U = k sigma_tot of the A-weighted level (Eq. 23), in decibels.

With sigma_omc = 2,0 dB and k = 2 this is the 5 dB of the 9.5 EXAMPLE.

HardWalledSoundPowerResult.plot(
    ax: Axes | None = None,
    *,
    language: str = 'en',
    **kwargs: Any,
) -> Axes

Plot the determined spectrum with the A-weighted total annotated.

One bar per octave band of LW (or LJ), a band that 8.1.3 makes an upper bound hatched and named as one, a band that fails the background requirement of 4.5 otherwise cross-hatched and named as that, and the expanded uncertainty as an error bar where sigma_omc was given. Requires matplotlib (pip install phonometry[plot]).

Parameters

Returns: The axes.

HardWalledSoundPowerResult.report(
    path: str,
    *,
    metadata: ReportMetadata | None = None,
    engine: str = 'reportlab',
    verbose: bool = False,
    language: str = 'en',
) -> str

Render the ISO 3743-1 determination as a one-page test sheet.

The sheet states the method and its accuracy grade (the comparison with a reference sound source in a hard-walled test room, ISO 3743-1:2010, grade 2), an optional metadata header, the per-band table of the mean level of the source, the mean level and the calibrated sound power level of the reference sound source and the determined LW (or LJ), the spectrum, the boxed A-weighted level with the expanded uncertainty and its coverage factor, an optional verdict against a declared limit, and a basis strip with the Eq. 14 (or Eq. 20) chain and its corrections. A band that is an upper bound is marked * and named one beneath the table, with the background requirement not fulfilled, which 8.1.3 asks a report to state in its tables; any other band whose background requirement was not met is marked † and named as not fulfilling it.

Parameters

Returns: The written path as a str.

Raises

property

${\sigma }_{\mathrm{tot}}=\sqrt{{\sigma }_{R0}^2+{\sigma }_{\mathrm{omc}}^2}$ per band (Eq. 22), in decibels; NaN without sigma_omc and at 63 Hz.

property

Eq. (22) with the A-weighted row of Table 3, in decibels.

property

The A-weighted total of sound_energy_level_ref (Eq. B.2).

property

LJ + C2 under the reference meteorological conditions (Eq. A.3); NaN for a power determination.

property

The A-weighted total of sound_power_level_ref.

C2 is the same in every band, so it passes through Eq. (B.1) unchanged: the total moves by C2 too.

property

LW + C2 under the reference meteorological conditions (Eq. A.1); NaN for an energy determination.

property

Per band, whether 8.1.3 calls the level an upper bound.

The margin of the source under test was measured and fell below 6 dB while the reference source’s margin met it.

reproducibility_from_round_robin(
    total_db: float,
    operating_db: float,
) -> float

The standard deviation of reproducibility of the method from a round robin test (ISO 3743-1:2010 Eq. 24).

A round robin gives the total standard deviation ${\sigma }_{\mathrm{tot}}^{\prime }$ of one source, which still carries the instability of that source’s operating and mounting conditions; taking it out in quadrature leaves what the method itself contributes. ISO 3743-2:2018 prints the same relation as its Formula (14) with a plus sign, which would make the method’s share larger than the total it is extracted from; the sentence after it (the result is imprecise when ${\sigma }_{\mathrm{tot}}$ is only slightly higher than ${\sigma }_{\mathrm{omc}}$) holds only for the difference, and the library evaluates the difference for both parts (see docs/ERRATA.md).

The clause adds a condition on the inputs: to keep the result from being a small number of low accuracy, ${\sigma }_{\mathrm{omc}}$ should not exceed ${\sigma }_{\mathrm{tot}}/\sqrt{2}$. Beyond it this warns; the value is still returned.

Parameters

Returns: ${\sigma }_{R0}^{\prime }$, in decibels.

Raises

sound_energy_hard_walled(
    event_levels: ArrayLike,
    levels_ref: ArrayLike,
    lw_ref: ArrayLike | ReferenceSourceCalibration,
    frequencies: ArrayLike,
    *,
    events: int | None = None,
    background_levels: ArrayLike | None = None,
    integration_time_s: float | None = None,
    background_levels_ref: ArrayLike | None = None,
    temperature_c: float = 23.0,
    static_pressure_kpa: float = 101.325,
    sigma_omc_db: float | None = None,
    coverage_factor: float = 2.0,
) -> HardWalledSoundPowerResult

Sound energy level of a source emitting bursts, in a hard-walled test room, by comparison with a reference sound source (ISO 3743-1:2010, 8.2).

The single event levels come in one of the two forms 7.6 admits. Measured one event at a time, the $N_{\mathrm{e}}$ events are on the first axis and Eq. (15) takes their energy mean; measured once over $N_{\mathrm{e}}$ successive events, Eq. (16) subtracts $10{\log }_{10}{N}_{\mathrm{e}}$. The level of one event is then averaged over the source locations (Eq. 17) and the positions (Eq. 18), corrected by the rule of 8.2.3 around Eq. (19), and

The reference source is steady and measured time-averaged over 30 s (7.6), so $K_{1(\mathrm{RSS})}$ follows 8.1.3 unchanged. For a source steady over the interval $T$ this gives $L_J=L_W+10{\log }_{10}(T/T_0)$, $T_0$ = 1 s (3.4 NOTE 1).

Parameters

Returns: HardWalledSoundPowerResult with quantity='energy'.

Raises

sound_power_hard_walled(
    levels: ArrayLike,
    levels_ref: ArrayLike,
    lw_ref: ArrayLike | ReferenceSourceCalibration,
    frequencies: ArrayLike,
    *,
    background_levels: ArrayLike | None = None,
    background_levels_ref: ArrayLike | None = None,
    temperature_c: float = 23.0,
    static_pressure_kpa: float = 101.325,
    sigma_omc_db: float | None = None,
    coverage_factor: float = 2.0,
) -> HardWalledSoundPowerResult

Sound power level of a small movable source in a hard-walled test room, by comparison with a reference sound source (ISO 3743-1:2010, 8.1).

The octave-band time-averaged levels of the source under test are energy-averaged over the source locations, if the survey of 7.4 asked for more than one (Eq. 9), and over the microphone positions (Eq. 10); those of the reference source, placed where the source under test stood (7.2), over the same positions (Eq. 11), and the background likewise (Eq. 12). Each mean is corrected for background noise by the rule of 8.1.3 around Eq. (13), and

at the meteorological conditions of the test; sound_power_level_ref adds the Annex A correction and sound_power_level_a is the Annex B total.

Parameters

Returns: HardWalledSoundPowerResult with quantity='power'.

Raises

SourceLocationPlan(
    standard: str,
    frequencies: np.ndarray,
    standard_deviation_db: np.ndarray,
    source_locations: np.ndarray,
    additional_room_locations: np.ndarray,
    microphone_positions: int,
    a_weighted_standard_deviation_db: float = nan,
    a_weighted_source_locations: int = 0,
)

How many source locations a determination needs, from a preliminary survey of the room (ISO 3743-1:2010 7.4, Table 2; ISO 3743-2:2018 9.4, Table 3).

standard_deviation_db is the estimated standard deviation $s_{\mathrm{M}}$ of the preliminary levels per band, and source_locations the minimum number $N_{\mathrm{S}}$ of source locations its table asks for, in the same room. Part 1 sends a band above 4 dB to a second room as well: additional_room_locations counts the locations there, two in such a band and zero elsewhere, and always zero for Part 2. microphone_positions is the $N_{\mathrm{M}}$ the numbers are read for: the preliminary positions of Part 1, the column of Part 2’s Table 3.

spectral_character is Part 2’s 9.5 reading of $s_{\mathrm{M}}$ per band, 'broadband', 'narrow-band' or 'discrete tone', and None for Part 1, which draws no such conclusion. The a_weighted_... fields are the same for the A-weighted row of Part 2 when A-weighted levels were surveyed, NaN, 0 and None otherwise. The two spectral characters are read from the deviations and the 2,3 dB and 4 dB of 9.5, so they are read-only properties and not fields.

property

The 9.5 reading of the A-weighted deviation, or None without one.

property

Whether a band of Part 1 sends the determination to a second room.

SourceLocationPlan.plot(
    ax: Axes | None = None,
    *,
    language: str = 'en',
    **kwargs: Any,
) -> Axes

Plot $s_{\mathrm{M}}$ per band against the table’s class limits, with the number of source locations over each bar.

Requires matplotlib (pip install phonometry[plot]).

Parameters

Returns: The axes.

property

The largest $N_{\mathrm{S}}$ any band (or the A-weighted row) asks for in the test room, which is the number a single determination covering them all has to use.

property

Part 2’s 9.5 reading of $s_{\mathrm{M}}$ per band, or None for Part 1.

'broadband' below 2,3 dB, 'narrow-band' up to 4 dB and 'discrete tone' above.