Generated by dspark_conformance --metrics (48 kHz, stereo, block 512).

THD+N: spectral energy outside the fundamental, 1 kHz tone at -6 dBFS, relative to the fundamental, 20 Hz-20 kHz.
Noise floor: output RMS (dBFS) with silent input.
Spurious: worst non-harmonic component (dB re fundamental) with a 10.1 kHz tone at -6 dBFS - aliasing shows up here.
Latency: as reported by getLatency() (samples @ 48 kHz). Validated by the PDC null tests in this same suite.

Processor	Settings	Latency	THD+N @1 kHz	Noise floor	Spurious @10.1 kHz	Notes
Gain	-3 dB	0	<0.001 %	-240.0 dBFS	-153.8 dB
FilterEngine	LP 16 kHz, 24 dB/oct	0	<0.001 %	-240.0 dBFS	-153.5 dB
Equalizer (IIR)	+3 dB bell @ 500 Hz	0	<0.001 %	-240.0 dBFS	-153.6 dB
Equalizer (linear phase)	+3 dB bell @ 500 Hz	512	<0.001 %	-240.0 dBFS	-149.9 dB
Compressor	-20 dB, 4:1	0	0.057 %	-240.0 dBFS	-82.4 dB	gain-riding on steady tone
Limiter	-1 dBTP ceiling	96	<0.001 %	-240.0 dBFS	-153.6 dB
NoiseGate	-60 dB (open)	0	<0.001 %	-240.0 dBFS	-153.6 dB
Expander	-60 dB (inactive)	0	<0.001 %	-240.0 dBFS	-153.6 dB
DynamicEQ	3 kHz band, -30 dB, 3:1	0	<0.001 %	-240.0 dBFS	-153.3 dB
MultibandCompressor	3 bands, defaults	0	0.044 %	-240.0 dBFS	-82.4 dB
TransientDesigner	attack +20%	0	<0.001 %	-240.0 dBFS	-125.3 dB
DeEsser	7 kHz, -30 dB	0	0.006 %	-240.0 dBFS	-153.5 dB
Saturation (SoftClip)	drive 0 dB, 2x OS	64	1.962 %	-240.0 dBFS	-124.2 dB
Clipper (Analog)	0 dB in, 4x OS	96	1.058 %	-240.0 dBFS	-129.3 dB
TapeMachine	drive 0 dB, wow/flutter off	223	0.601 %	-240.0 dBFS	-99.4 dB
TubePreamp	1 stage, drive 0 dB	71	3.306 %	-240.0 dBFS	-121.2 dB
TransformerModel	drive 0 dB	64	0.635 %	-240.0 dBFS	-88.8 dB
Chorus	0.8 Hz, mix 50%	0	47.791 %	-240.0 dBFS	-9.3 dB	modulated: sidebands are the effect
Phaser	0.5 Hz	0	0.001 %	-240.0 dBFS	-101.7 dB	modulated: sidebands are the effect
Tremolo	5 Hz, depth 0.7	0	<0.001 %	-240.0 dBFS	-101.9 dB	amplitude modulation is the effect
Vibrato	5 Hz, depth 0.4	0	n/a	-240.0 dBFS	n/a	FM: spectrum displaced by design
RingModulator	300 Hz carrier	0	n/a	-240.0 dBFS	n/a	spectrum displaced by design
FrequencyShifter	+50 Hz	95	n/a	-240.0 dBFS	n/a	spectrum displaced by design
PitchShifter	+4 st	5184	n/a	-240.0 dBFS	n/a	spectrum displaced by design
TimeStretch	ratio 1.081 (-8% tempo)	5632	n/a	-240.0 dBFS	n/a	stretched stream: block counts do not carry duration
GranularProcessor	density 30/s	0	39.652 %	-240.0 dBFS	-10.2 dB	granular texture is the effect
SpectralDenoiser	reduction 12 dB	2048	<0.001 %	-240.0 dBFS	-150.9 dB
SpectralFreeze	live path, N 2048, hop 512	2048	<0.001 %	-240.0 dBFS	-150.9 dB	hold engages on request; the live path is measured
PitchCorrector	chromatic, hard snap	4096	n/a	-240.0 dBFS	n/a	retunes the probe tone to the nearest semitone: displaced by design
AlgorithmicReverb	Hall, mix 40%	0	0.083 %	-240.0 dBFS	-47.1 dB	time-based: tail energy
ConvolutionReverb	synthetic IR, mix 30%	512	<0.001 %	-240.0 dBFS	-136.6 dB	time-based: tail energy
StereoWidth	width 1.4, bass mono 120 Hz	0	<0.001 %	-240.0 dBFS	-153.6 dB
Panner	pan 0.4 R	0	<0.001 %	-240.0 dBFS	-153.6 dB
DCBlocker	10 Hz, order 4	0	<0.001 %	-240.0 dBFS	-153.7 dB

Modulated, inharmonic-by-design and time-based processors necessarily show high "THD+N" - for them the column documents the character at the listed settings rather than a defect. Linear processors read at the measurement's own floor.

Analysis accuracy

Measured on synthetic material with exactly known answers, at 44.1 kHz, by this same run.

Readout	Material	Quantity	Result
`BeatTracker` tempo	click track, 40 to 240 BPM	worst tempo error over the sweep	-0.014 BPM (at 210 BPM)
`BeatTracker` grid	same	worst beat placement error	5.56 ms (at 45 BPM)
`BeatTracker` grid	same	beats missed / invented	0 / 0
`BeatTracker` confidence	same	lowest reported over the sweep	0.993
`LoopFinder` endpoints	three-tone texture, planted 4869-sample loop	endpoint recovery	exact
`LoopFinder` seam cost	same	reported normalized cost at the planted splice	0.00e+00
`PitchCorrector` tuning	harmonic tones 45 cents off five notes, chromatic scale	worst steady-state error	+0.87 cents (note 76)

The tempo figures are the WORST point of the sweep with its sign, not an average and not an endpoint: a tracker's error is not monotonic in tempo, so an average would hide the tempo it is worst at. Beats are matched one-to-one inside 70 ms.

Rate conversion, sample playback and low-frequency dynamics

Measured in double precision by this same run. THD+N is the residual of a least-squares fit of the tone, relative to the tone; alias is the output energy of a tone that has no place below the output Nyquist frequency, relative to its input energy.

Component	Settings	Quantity	Result
`Resampler`	Draft, 44.1 to 48 kHz	THD+N, 1 kHz	-77.3 dB
`Resampler`	Draft, 48 to 44.1 kHz	alias of a 23 kHz tone	-61.3 dB
`Resampler`	Normal, 44.1 to 48 kHz	THD+N, 1 kHz	-120.5 dB
`Resampler`	Normal, 48 to 44.1 kHz	alias of a 23 kHz tone	-104.4 dB
`Resampler`	High, 44.1 to 48 kHz	THD+N, 1 kHz	-161.3 dB
`Resampler`	High, 48 to 44.1 kHz	alias of a 23 kHz tone	-146.0 dB
`Resampler`	Ultra, 44.1 to 48 kHz	THD+N, 1 kHz	-237.1 dB
`Resampler`	Ultra, 48 to 44.1 kHz	alias of a 23 kHz tone	-219.6 dB
`Sampler`	1 kHz recording, -12 semitones	THD+N	-117.6 dB
`Sampler`	1 kHz recording, +7 semitones	THD+N	-123.4 dB
`Sampler`	1 kHz recording, +19 semitones	THD+N	-129.9 dB
`Sampler`	18 kHz recording, +12 semitones	alias	-110.2 dB
`Compressor`	Peak detector, 30 Hz 14 dB over 4:1, 5/100 ms	THD+N	-35.1 dB
`Compressor`	Hilbert detector, 30 Hz 14 dB over 4:1, 5/100 ms	THD+N	-96.3 dB