A transparent comparison of our DSP engines. No made-up quality scores.
Deterministic synthetic signals, 12 seconds, 24 kHz stereo. Median of three warmed renders on this host. Sun/Moon use their model renderers; Classic uses its Governor DSP chain with fixed settings, excluding the automatic analysis/proposal stage. Current V9.1 models only. Excludes network, queue and file decoding. BS.1770 integrated LUFS; four-times FIR true-peak estimate. No listener study or overall quality score.
Run python scripts/benchmark_engines.py. Generated 2026-10-02T04:37:43.376775+00:00. Engine 1.0.0.
Windows-11-10.0.26200-SP0 · AMD64 Family 25 Model 97 Stepping 2, AuthenticAMD · 16 logical CPUs
Python 3.12.14 · NumPy 2.5.3 · SciPy 1.18.1
DSP source SHA-256: 518c92c4a923367e7f974484f8fe6c20afa9568c4231880b3ccc6e3827ba59c0
Download raw measurements ↗These are synthetic tests, not released songs or a listening study. Faster does not mean better sound. A lower crest factor means denser dynamics; a higher one retains more peak contrast. True peak is a 4× estimate. Loudness can fall below the target when peak protection needs headroom.
The result comes first. Compare sonic character, finishing controls and matching audio — then the credit cost.
Choose Sun for the most extensive processing and delivery options.
Highest measured true peak -1.0 dBTP
Mean loudness target error 0.09 LU
Choose Moon for a warmer Pro finish at a lower credit cost.
Highest measured true peak -1.102 dBTP
Mean loudness target error 0.0 LU
Choose Classic for the established chain without a Pro subscription.
Highest measured true peak -0.9 dBTP
Mean loudness target error 0.02 LU
Its extra credits pay for spectral repair, separate band dynamics, exclusive protection tools and expanded delivery. Moon offers a different warmer character. The preferred sound depends on your track: listen above, then compare the matching measurements below.
Lower means fewer credits spent.
Same budget, different output counts.
Credit arithmetic only. Sun and Moon require Pro; its weekly allowance is 2,000 credits.Mean of the six synthetic test medians. Lower is faster.
Development host, warmed DSP only; excludes upload, queue, decoding and analysis.Sun includes the full set; Moon includes five.
These are the seven new finishing tools, not a count of every processing stage. Core mastering is available in all three.Compare cost against measurable aspects of the audio result, or processing speed.
Audio metrics describe different aspects of sound, not an overall listening score. More transient contrast can preserve punch; tighter loudness matching and peak protection serve different goals. Sun adds spectral repair and exclusive finishing controls that these meters cannot rate. Use the matching audio samples to judge the finish.Choose Moon for credit efficiency, Sun for its additional controls and delivery options, or Classic for core mastering without a Pro subscription. Credit prices exclude the subscription cost. Listen to the matching samples below to judge the sound.
Aggregate of the measured synthetic tests below. Mean of per-test median render times; no subjective sound score.
| Model | Mean render | Mean target error | Highest true peak | Mean crest |
|---|---|---|---|---|
| Sun | 279.99 ms | 0.09 LU from −14 | -1.0 dBTP | 9.11 dB |
| Moon | 200.1 ms | 0.0 LU from −14 | -1.102 dBTP | 8.36 dB |
| Classic | 171.48 ms | 0.02 LU from −14 | -0.9 dBTP | 9.34 dB |
| Sun · 20 credits | Moon · 5 credits | Classic · 10 credits | |
|---|---|---|---|
| Core dynamics | Three independently controlled bands | Linked broadband glue | Governor chain |
| Tone | Adaptive spectral repair | Oversampled harmonic warmth | Manual EQ and tone |
| Custom profiles | Yes · Pro | Yes · Pro | — |
| New finishing tools | All seven | Five | — |
| Exclusive delivery | Frequency layers; Spatial / ADM | Standard stereo | Standard stereo |
| Tradeoff | More control can reduce peak contrast | Deliberate coloration; less transparent | No custom profile or flagship stages |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -17.354 | -3.534 dBTP | 14.647 dB | 0.9987 | |
| Orynos V9.1 Sun | 280.46 ms | -14.541 | -1.0 dBTP | 14.668 dB | 0.9974 | |
| Orynos V9.1 Moon | 200.6 ms | -14.0 | -1.102 dBTP | 13.333 dB | 0.999 | |
| Orynos V9.1 Classic | 220.89 ms | -14.084 | -0.9 dBTP | 14.129 dB | 0.9988 |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -21.557 | -15.704 dBTP | 7.239 dB | 0.9996 | |
| Orynos V9.1 Sun | 279.18 ms | -14.0 | -7.244 dBTP | 8.368 dB | 0.9993 | |
| Orynos V9.1 Moon | 200.3 ms | -14.0 | -8.003 dBTP | 7.354 dB | 0.9997 | |
| Orynos V9.1 Classic | 164.08 ms | -14.007 | -8.206 dBTP | 7.299 dB | 0.9996 |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -17.934 | -10.004 dBTP | 10.796 dB | 0.9262 | |
| Orynos V9.1 Sun | 279.96 ms | -14.0 | -5.961 dBTP | 11.097 dB | 0.9041 | |
| Orynos V9.1 Moon | 200.58 ms | -14.0 | -6.366 dBTP | 10.459 dB | 0.9455 | |
| Orynos V9.1 Classic | 164.51 ms | -14.037 | -5.255 dBTP | 11.906 dB | 0.9304 |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -17.383 | -13.003 dBTP | 6.632 dB | 0.9992 | |
| Orynos V9.1 Sun | 286.12 ms | -14.0 | -9.258 dBTP | 7.153 dB | 0.9989 | |
| Orynos V9.1 Moon | 201.64 ms | -14.0 | -9.646 dBTP | 6.465 dB | 0.9995 | |
| Orynos V9.1 Classic | 162.14 ms | -14.0 | -9.634 dBTP | 6.621 dB | 0.9993 |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -19.156 | -16.678 dBTP | 6.066 dB | 1.0 | |
| Orynos V9.1 Sun | 277.02 ms | -14.0 | -11.24 dBTP | 7.06 dB | 0.9995 | |
| Orynos V9.1 Moon | 198.17 ms | -14.0 | -11.176 dBTP | 6.603 dB | 1.0 | |
| Orynos V9.1 Classic | 158.7 ms | -13.999 | -9.119 dBTP | 9.157 dB | 1.0 |
| Engine | Median render | LUFS | True peak | Crest factor | Stereo correlation | Listen |
|---|---|---|---|---|---|---|
| Original | — | -17.64 | -12.237 dBTP | 5.735 dB | 0.3989 | |
| Orynos V9.1 Sun | 277.21 ms | -14.0 | -8.847 dBTP | 6.336 dB | 0.5535 | |
| Orynos V9.1 Moon | 199.31 ms | -14.0 | -8.978 dBTP | 5.965 dB | 0.4562 | |
| Orynos V9.1 Classic | 158.54 ms | -13.999 | -8.131 dBTP | 6.926 dB | 0.3994 |
Spectral balance, independent band dynamics and bass stereo stabilisation. More control, but conservative compression can reduce peak contrast.
Linked dynamics and oversampled harmonics for body and warmth. Fewer credits; deliberate coloration is less suitable for transparent archival work.
Classic retains the Governor chain. These measurements describe signal behaviour, not a subjective quality ranking.
| Provider | Status | Evidence |
|---|---|---|
| LANDR | Not measured | No matching source/export supplied. |
| BandLab | Not measured | No matching source/export supplied. |
Upload your original and the corresponding provider masters below. We render Sun, Moon and Classic from that same original at −14 LUFS with a −1 dBTP ceiling, then measure our renders and your provider files with the same meters. Provider strengths and weaknesses cannot be established without comparable audio and listening evidence.
OrynosMusic stays available whether or not you contribute.