broadcast.quasi_peak

Psophometric quasi-peak programme-level meter (ITU-R BS.468-4 clause 2).

ITU-R BS.468-4 specifies the instrument that measures audio-frequency noise voltage in sound broadcasting as two parts. Clause 1 is the weighting network, realised elsewhere in this library (phonometry.filters.weighting_filter with curve="468", from the Fig. 1a ladder). Clause 2 is the detector this module implements: a quasi-peak value method, whose reading is reported in dBqps (clause 3).

Clause 2 prints no dynamics. It gives no time constant, no rise time, no decay law, no transfer function and no differential equation. The words “time constant” occur once in the whole Recommendation, in an informative Note that offers “two peak rectifier circuits of different time-constants connected in tandem” as a possible arrangement after full-wave rectification, and the preamble says the performance “may be realized in a variety of ways”. What clause 2 does give is eleven acceptance windows: Table 2 reads a single 5 kHz tone burst at eight durations from 1 ms to 200 ms, Table 3 reads a train of 5 ms bursts at 2, 10 and 100 per second, and each cell carries a reference reading with a lower and an upper limit, expressed as a percentage of the reading the same tone gives steadily. Those eleven windows are the entire specification of the dynamics, and they are what verify_quasi_peak_dynamics checks.

The one absolute statement clause 2 does make is clause 2.6: a steady 1 kHz sine at 0.775 V r.m.s. shall read 0.775 V, that is 0 dBqps. So the detector reads the r.m.s. of a steady sine, not its peak, and that fixes the scale of the whole instrument (NOTE 2 to clause 1: “The whole instrument is calibrated at 1 kHz”). Here that scale is not a stored number but a measurement: quasi_peak_meter runs its own chain on a 1 kHz sine of 1 V r.m.s. at the caller’s sample rate and divides by what it reads, so clause 2.6 holds exactly at every rate rather than at the one rate a frozen constant would have been fitted at.

Six decisions, none of them in BS.468-4, all of them inseparable: change any one and BS468_BALLISTICS has to be refitted.

• Full-wave rectification as $|x|$, following the Note’s own wording. An analytic-signal envelope also meets all eleven windows when it is refitted from scratch, with a three times larger worst-case residual.
• A peak rectifier as a one-pole recursion with different rise and fall rates, the same asymmetric kernel the impulse time weighting uses (phonometry.filters.weighting.time_weighting), discharging toward the input rather than toward zero. Refitting toward zero instead moves the worst residual by 0.002 dB, so the discharge law is not load-bearing.
• One symmetric reading device after it, first order. Clause 2.5 treats “the reading device” as a separate object, and a peak rectifier alone cannot meet the tables: its best fit is 4.3 dB out at the worst point and leaves only 4 of the 11 readings inside their windows.
• The reading is the maximum of the reading device’s output over the record, which is what an operator writes down from a pointer. The record therefore has to include the decay: a burst that ends at the last sample is read before the needle has finished rising.
• The reference for Tables 2 and 3 is the same chain fed a steady 5 kHz tone through the same network, so both tables are ratios and every absolute factor, the network’s +11.7 dB at 5 kHz included, cancels.
• The ballistics run at the caller’s rate on the record phonometry.filters.weighting_filter returns. Measured across 32, 44.1, 48, 96 and 192 kHz, no reading moves by more than 0.080 dB, or 4.2 % of its own window.

The tables use one carrier, one waveform, one amplitude, eight durations and three repetition rates, and that is the whole specification. There is no reference implementation, no tabulated impulse response and no worked example in BS.468-4, in IEC 60268-1:1985 Appendix A or in either 1988 amendment. So the response to noise, which is the instrument’s entire purpose, is unverifiable: two conforming detectors will disagree on noise, on speech and on programme material by an amount nothing in the document bounds. So is the relationship between a dB(468) quasi-peak figure and a dB(A) r.m.s. one, and so is everything outside the tabulated stimuli (bursts shorter than 0.6 ms or longer than 200 ms, carriers other than 5 kHz, anything that is not a gated sine).

Four clauses are computable and pass by construction rather than by measurement, because rectification, first-order recursions and a maximum are all positively homogeneous and cannot overshoot a monotone step: clause 2.1’s two attenuator variants, clause 2.3’s +-1 dB over 20 dB of 0.6 ms bursts, clause 2.4’s 0.5 dB reversibility ($|-x|=|x|$, so the difference is exactly zero) and clause 2.5’s 0.3 dB overswing cap. They are properties of the arithmetic, not evidence about the design. The rest of clauses 2.3, 2.6 and 2.7 has nothing to compute at all: overload capacity above full scale, the law of a logarithmic stage, a calibrated scale range of at least 20 dB, an input impedance of at least 20 kOhm and a 600 Ohm termination are instrument hardware, and “80 % of full scale” constrains where a real needle sits during the test rather than the quantity, which is why this module does not model a full scale at all.

There is no report fiche. BS.468-4 prescribes no report format, no declaration form and no worked example; the conformity evidence is the eleven windows, and its home is docs/CONFORMANCE.md through verify_quasi_peak_dynamics.

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

Constant (phonometry.broadcast.quasi_peak.QuasiPeakBallistics).

Constant (float).

DBQPS_REFERENCE = 0.775

quasi_peak_meter(
    x: SignalInput,
    fs: float | None = None,
    *,
    weighted: bool = True,
    reference: float | None = None,
) -> QuasiPeakResult

Read a record with the ITU-R BS.468-4 quasi-peak meter.

Runs clause 2’s detector over the whole record and returns the largest excursion of the reading device, on the scale clause 2.6 fixes. The reading rule is the maximum over the record, so the record must include the decay of whatever it is measuring: a burst that ends at the last sample is read before the needle has finished rising.

The chain and everything it chooses that the Recommendation does not are described in the module docstring; the time constants are BS468_BALLISTICS, a fit to Tables 2 and 3 rather than a quoted value, and verify_quasi_peak_dynamics is what checks them.

Units. The detector itself is unit-agnostic: rectification, first-order recursions and a maximum carry whatever unit the record arrived in, and the eleven acceptance windows are ratios in which every unit cancels. Only the level needs a unit, and clause 2.6 gives BS.468-4 exactly one, the volt. A bare array or an uncalibrated Signal is therefore read against 0.775 V and its level_db is dBqps. A calibrated Signal is analysed in pascals, as everywhere else in this library, and then has to be given a reference of its own: 0.775 V is not a pressure, and BS.468-4 offers no pressure to put in its place. result.level_unit names whichever scale came out.

Parameters

Returns: A QuasiPeakResult.

Raises

QuasiPeakBallistics(charge: float, discharge: float, reading_device: float)

Time constants of the quasi-peak chain, in seconds.

Attributes

QuasiPeakResult(
    reading: float,
    level_db: float,
    reference: float,
    trace: np.ndarray,
    fs: float,
    weighted: bool,
)

A quasi-peak reading of one record (ITU-R BS.468-4 clause 2).

Attributes

property

The name of the scale level_db is on.

The one place that answers whether the level may be called dBqps, which clause 3 defines only against the 0.775 V of clause 2.6. Anything else is named by its own reference instead, because BS.468-4 gives no other scale a name — a quasi-peak sound pressure level read against 20 uPa is a perfectly good number and is not dBqps.

The name is read off the reference, which is the only unit-bearing thing the result holds: the samples arrive in whatever unit they were already in and the library never learns which. A caller who refers a voltage record to 20 uV rather than to 20 uPa gets the pascal name; nothing here can tell those two apart, and no other reference is close enough to either to collide.

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

Plot the quasi-peak trace over time, with the reading marked.

Requires matplotlib (pip install phonometry[plot]); returns the Axes.

Parameters

Returns: The axes.

property

Time of each sample of trace, in seconds.

verify_quasi_peak_dynamics(
    fs: float = 48000.0,
    ballistics: QuasiPeakBallistics = ...,
) -> dict[str, Any]

Check the detector against the eleven acceptance windows of clause 2.

Runs the clause 2.1 and 2.2 stimuli exactly as they are specified: a 5 kHz sine starting at a zero crossing and lasting an integral number of full periods (at 5 kHz one period is 0.2 ms, so the eight tabulated durations are 5, 10, 25, 50, 100, 250, 500 and 1000 cycles), through the weighting network, and reads each one against the same chain’s steady reading for the same tone. Table 3 repeats the 5 ms burst at 2, 10 and 100 per second.

This is the whole conformance statement BS.468-4 authorises for a detector, and it is an acceptance region rather than a set of values: every reading inside a window conforms, and the Recommendation tolerates about 2.5 dB of disagreement between two conforming instruments on a single burst. deviation_db is reported against the printed reference reading as well, but that column is a self-imposed regression bound, not conformance: the reference is printed to two significant figures on nine of the eleven cells, so its own quantum is 0.027 to 0.108 dB.

The stimulus amplitude does not appear anywhere: both tables are ratios to the steady reading at the same amplitude, and a chain built from rectification, first-order recursions and a maximum is exactly homogeneous, so clause 2.1’s two attenuator variants give the same answer by construction.

Parameters

Returns: Dict with fs, passed (every reading inside its window), worst_margin_db (the smallest margin over the eleven, negative when one is outside), worst_deviation_db (the largest departure from a printed reference reading) and stimuli: eleven rows of {"stimulus", "table", "reading_percent", "lower_percent", "reference_percent", "upper_percent", "deviation_db", "margin_db"}.

Raises