AudioAtlas Report - audioatlas

How to read this report

Start with Key metrics for level and headroom context, scan Findings for checks worth prioritizing, then inspect the relevant plots and verify by listening.

This report provides descriptive context, not professional mastering approval or a universal pass/fail result.

Relative dB values use this track's strongest measured content as 0 dB. They show shape within this song and are not calibrated dBFS.

Check before delivery / worth a listen / for reference indicate priority, not quality.

A report can have no prioritized findings; the plots still describe the track's measured shape.

Key metrics

-13.125

Integrated LUFS

LUFS

-0.439

True peak

dBTP

-0.458

Sample peak

dBFS

-17.437

RMS

dBFS

12.686

PLR

Clipped samples

Near-clipping samples

0.800

Median stereo correlation

Pearson r

-9.003

Median side/mid ratio

Findings

Findings are measurement-based observations derived from the analysis. They highlight values or regions worth checking by ear; they are not quality judgments.

for reference stereo

Stereo field shows sustained low-correlation / side-heavy regions

Evidence:

Minimum frame correlation: -0.473.
Total time below 0 correlation: 0.511 seconds across 2 region(s).
Total time below 0.3 correlation: 1.115 seconds across 3 region(s).

Why it matters: Low correlation and higher side energy can point to passages where mono playback may change tone, level, apparent width, or center focus.

Does not mean: This does not mean the stereo image is incorrect; wide and phase-rich material can be intentional.

Suggested listening checks

Inspect the stereo correlation timeline around the lowest-correlation regions.
Listen in mono around sustained low-correlation regions if mono compatibility matters.
Inspect the mid/side energy plot around side-heavy regions.

3 regions, total 1.115s, longest 0.511s.

47.067s-47.415s
54.567s-54.822s
70.310s-70.821s

Plots

Visual maps generated from the analysis.

Waveform + RMS Envelope

waveform_rms.png

What this shows: raw samples with the RMS envelope overlaid. It is amplitude context, not a quality judgment.

Frame RMS Timeline

rms_timeline.png

What this shows: frame-by-frame RMS energy over time. It describes energy movement within this track, not a loudness target.

Sample Histogram

sample_histogram.png

What this shows: sample-value distribution with clipping and near-clipping thresholds. Threshold markers do not prove audible distortion.

Log-Frequency Spectrogram

log_spectrogram.png

What this shows: frequency content over time on a log-frequency axis. Relative dB values use this track's strongest measured content as 0 dB. They show shape within this song and are not calibrated dBFS.

Understanding these numbers

Metric glossary

LUFS

Integrated LUFS is a whole-track loudness measurement weighted toward human hearing. It gives delivery context, not a quality judgment.

Short-term LUFS

Short-term LUFS is a time-varying K-weighted loudness measurement using 3 s windows. It shows where this track is louder or quieter over time and is distinct from RMS.

True peak

True peak estimates reconstructed peaks between samples. Values above 0 dBTP can matter for playback, conversion, or encoding headroom.

Sample peak

Sample peak is the largest stored sample value in the file. It is not loudness or density.

RMS

RMS is average signal energy. The RMS timeline is useful for seeing where energy rises or falls across the track.

Crest factor

Crest factor is peak-to-RMS contrast in dB. It describes measured peak contrast, not punch, quality, or dynamic range.

PLR

PLR is approximate true peak minus integrated loudness. It describes the measured distance between those two values; it does not by itself identify compression, transient quality, or a delivery problem.

Clipping / near-clipping

Clipping counts samples at the configured ceiling; near-clipping counts samples close to it. Use the waveform and histogram to inspect where these samples occur.

Stereo correlation

Stereo correlation describes the L/R relationship. +1 means nearly identical channels; 0 means loosely related; negative values indicate opposite-polarity/decorrelated content.

Side/mid ratio

Side/mid ratio compares stereo-difference energy with center energy. 0 dB means side and mid energy are similar; more negative means mid-dominant; closer to 0 means more side energy.

Spectral centroid

Spectral centroid is the spectrum's center-of-mass frequency. It moves higher when energy shifts upward in frequency, and lower when energy is weighted toward lows/mids.

Rolloff

Spectral rolloff marks the frequency below which most measured spectral energy sits.

Spectral bandwidth

Spectral bandwidth describes how spread out the spectrum is around the centroid.

Average spectrum

Average spectrum is the long-term frequency profile of the track. Relative values show shape within the file.

Relative mean band power

This view averages spectral power per included FFT bin in each broad frequency band, then normalizes values within the file. It is not integrated total band energy and does not indicate absolute dBFS level.

Onset density

Onset density is an attack/activity map for this track. It is not punch, groove quality, drum hits per second, or mix quality. Absolute onset-density values are not reliable quality comparisons across songs.

Chroma CQT

Chroma CQT shows pitch-class energy within this track. It is not key detection, chord detection, or harmonic-quality analysis.

Relative dB

Relative dB plots show shape within this track. They are useful for shape within this track; not comparable to dBFS values from meters or other songs.

Technical details

Level metrics

duration_seconds	70.984
sample_rate	44100
channels	2
sample_peak_linear	0.949
sample_peak_dbfs	-0.458
true_peak_linear	0.951
true_peak_dbtp	-0.439
rms_linear	0.134
rms_dbfs	-17.437
crest_factor_db	16.979
integrated_lufs	-13.125
plr_db	12.686
clipped_samples	0
clipped_percent	0.000
near_clipping_samples	0
near_clipping_percent	0.000
dc_offset_per_channel	0.000, 0.000
peak_dbfs_per_channel	-0.458, -1.382
rms_dbfs_per_channel	-17.535, -17.341
true_peak_linear_per_channel	0.951, 0.853
true_peak_dbtp_per_channel	-0.439, -1.376

Crest factor timeline

frame_length	4096
hop_length	1024
frames	3058
crest_factor_db_min	5.422
crest_factor_db_median	10.583
crest_factor_db_max	22.800

Stereo metrics

frame_length	4096
hop_length	1024
frames	3058
defined_frames	3050
undefined_frames	8
correlation_min	-0.473
correlation_max	0.999
correlation_mean	0.746
correlation_median	0.800
overall_correlation	0.728
correlation_below_0_time_ranges	start: 5.921, end: 6.107, duration: 0.186, start: 16.672, end: 16.765, duration: 0.093, start: 47.137, end: 47.392, duration: 0.255, start: 53.267, end: 53.313, duration: 0.046, start: 54.590, end: 54.683, duration: 0.093, start: 62.531, end: 62.647, duration: 0.116, start: 70.356, end: 70.612, duration: 0.255
correlation_below_0_3_time_ranges	22 item(s)
warnings	one or more frames are below correlation_min_rms_dbfs; correlation is undefined

Spectrum metrics

nperseg	8192
bins	4097
strongest_bin_hz	64.600
strongest_bin_db	0.000
band_measurement	relative_mean_power_per_fft_bin
band_mean_power	sub: low_hz: 20.000, high_hz: 60.000, mean_power_db: -3.560, energy_db: -3.560, bass: low_hz: 60.000, high_hz: 120.000, mean_power_db: -2.108, energy_db: -2.108, low_mid: low_hz: 120.000, high_hz: 350.000, mean_power_db: -8.105, energy_db: -8.105, mid: low_hz: 350.000, high_hz: 2000.000, mean_power_db: -13.736, energy_db: -13.736, presence: low_hz: 2000.000, high_hz: 5000.000, mean_power_db: -19.835, energy_db: -19.835, high: low_hz: 5000.000, high_hz: 10000.000, mean_power_db: -27.270, energy_db: -27.270, air: low_hz: 10000.000, high_hz: 20000.000, mean_power_db: -36.817, energy_db: -36.817
highest_mean_power_band	bass
band_energies	sub: low_hz: 20.000, high_hz: 60.000, mean_power_db: -3.560, energy_db: -3.560, bass: low_hz: 60.000, high_hz: 120.000, mean_power_db: -2.108, energy_db: -2.108, low_mid: low_hz: 120.000, high_hz: 350.000, mean_power_db: -8.105, energy_db: -8.105, mid: low_hz: 350.000, high_hz: 2000.000, mean_power_db: -13.736, energy_db: -13.736, presence: low_hz: 2000.000, high_hz: 5000.000, mean_power_db: -19.835, energy_db: -19.835, high: low_hz: 5000.000, high_hz: 10000.000, mean_power_db: -27.270, energy_db: -27.270, air: low_hz: 10000.000, high_hz: 20000.000, mean_power_db: -36.817, energy_db: -36.817
strongest_band	bass

Spectral shape

n_fft	4096
hop_length	1024
frames	3058
valid_frames	3058
undefined_frames	0
centroid_mean_hz	3584.593
centroid_median_hz	3758.430
centroid_min_hz	624.737
centroid_max_hz	11085.554
rolloff_85_median_hz	8118.018
rolloff_95_median_hz	11660.229
bandwidth_median_hz	3816.067
centroid_elevated_threshold_hz	5637.646
centroid_reduced_threshold_hz	1879.215
centroid_large_shift_threshold_hz	2818.823
centroid_elevated_time_ranges	76 item(s)
centroid_reduced_time_ranges	66 item(s)
centroid_large_shift_time_ranges	13 item(s)

Relative mean band power timeline

measurement	relative_mean_power_per_fft_bin
n_fft	4096
hop_length	1024
frames	3058
valid_frames	3058
undefined_frames	0
band_names	sub, bass, low_mid, mid, presence, high, air
bands	sub: median_db: -36.521, mean_db: -42.613, max_db: -0.761, min_db: -100.000, elevated_threshold_db: -30.521, reduced_threshold_db: -48.521, elevated_time_ranges: 49 item(s), reduced_time_ranges: 42 item(s), bass: median_db: -24.023, mean_db: -34.126, max_db: 0.000, min_db: -100.000, elevated_threshold_db: -18.023, reduced_threshold_db: -36.023, elevated_time_ranges: 74 item(s), reduced_time_ranges: 21 item(s), low_mid: median_db: -24.461, mean_db: -27.491, max_db: -8.770, min_db: -100.000, elevated_threshold_db: -18.461, reduced_threshold_db: -36.461, elevated_time_ranges: 117 item(s), reduced_time_ranges: 55 item(s), mid: median_db: -29.394, mean_db: -31.039, max_db: -15.319, min_db: -100.000, elevated_threshold_db: -23.394, reduced_threshold_db: -41.394, elevated_time_ranges: 85 item(s), reduced_time_ranges: 58 item(s), presence: median_db: -35.632, mean_db: -44.293, max_db: -19.445, min_db: -100.000, elevated_threshold_db: -29.632, reduced_threshold_db: -47.632, elevated_time_ranges: 103 item(s), reduced_time_ranges: 50 item(s), high: median_db: -44.104, mean_db: -52.353, max_db: -26.706, min_db: -100.000, elevated_threshold_db: -38.104, reduced_threshold_db: -56.104, elevated_time_ranges: 90 item(s), reduced_time_ranges: 60 item(s), air: median_db: -54.293, mean_db: -61.318, max_db: -35.887, min_db: -100.000, elevated_threshold_db: -48.293, reduced_threshold_db: -66.293, elevated_time_ranges: 97 item(s), reduced_time_ranges: 69 item(s)
highest_mean_power_band_by_median	bass
strongest_band_by_median	bass

Onset density

hop_length	1024
frames	3058
smoothing_window_seconds	1.000
smoothing_window_frames	43
mel_bands	128
onset_strength_mean	1.298
onset_strength_median	0.722
onset_strength_max	23.542
onset_density_mean	1.295
onset_density_median	1.233
onset_density_max	2.530
high_onset_density_threshold	1.849
high_onset_density_time_ranges	21 item(s)
strongest_onset_density_time	17.067

Chroma CQT

frames	3058
hop_length	1024
pitch_classes	12 item(s)
mean_chroma	12 item(s)
dominant_pitch_class	C

Short-term LUFS

window_seconds	3.000
hop_seconds	0.100
frames	681
lufs_min	-29.823
lufs_median	-13.560
lufs_max	-9.928

Analysis config

n_fft	4096
hop_length	1024
window	hann
db_floor	-100.000
rms_frame_length	4096
clipping_threshold	0.999
near_clipping_threshold	0.990
true_peak_oversample	4
welch_nperseg	8192
max_plot_points	250000
correlation_min_rms_dbfs	-80.000
onset_density_window_seconds	1.000
short_term_lufs_window_seconds	3.000
short_term_lufs_hop_seconds	0.100
max_findings	8
band_finding_min_duration_seconds	0.500
band_finding_min_relative_db	-80.000
finding_min_time_range_seconds	0.250
report_max_time_ranges	8

Analysis provenance

format_version	1
audioatlas_version	0.2.0a8
summary_schema_version	0.2.1
finding_ruleset_version	0.2.0a2
analysis_config_sha256	25ae6d74ba8d101215c84b9d5316aea585c1c95f4cfe782873c701b1d804d689
measurement_code_sha256	2210006b8fa8d05456ce0f6bfd25c8df75e7b7f4d8c653f63adc5a68fbcd1299
finding_rule_code_sha256	236348735a92f4b6f4d5da87d0ac6aac09222def14533b5a8dc7a2cb8e3b6a16
dependencies	numpy: 2.4.6, scipy: 1.18.0, soundfile: 0.14.0, librosa: 0.11.0, numba: 0.65.1, pyloudnorm: 0.2.0
decoder	python_binding: soundfile, soundfile_version: 0.14.0, libsndfile_version: 1.2.2
measurement_methods	approximate_true_peak: method: scipy.signal.resample_poly, oversample_factor: 4, standards_grade_meter: False, integrated_loudness: method: pyloudnorm.Meter.integrated_loudness
environment	python_version: 3.13.14, python_implementation: CPython, platform_system: Linux, platform_machine: x86_64, byteorder: little
compatible_analysis_sha256	37a8fabf799300d774d4ee4901b4ad1d6ba61be195dc1adeaa61cf85a5e6d424
exact_environment_sha256	94c19d4e0482dac0d6056954c084255c4c2bf6bf9f803c5d1a122a47df039f35

Human notes

Notes stay in this browser for this report path. They are never written into the report files or sent over a network.

Notes are saved locally as you type.

Observations

EQ ideas

Dynamics notes

Stereo/image notes