PT · EN
Instrumental single by Nando Valle, 2:24, released on 11 September 2026. This page carries the measured sheet, the who the piece is dedicated to, and the originality check.
The rule of this house is the usual one: the track ships with its method. Here the method had one extra question to answer, and the question came from the title.
Also on Apple Music. Piano and orchestra, no guitar and no voice: it is the quietest piece in the catalogue, and the only one with nothing baroque about it.
The figures below come from the distributed file, measured on 11 September 2026
with ffmpeg 8.1.2, loudnorm filter (EBU R128):
| Quantity | Measured | Reference |
|---|---|---|
| Duration | 144.000 s | exactly 2:24 |
| Format | PCM 16-bit | 48 kHz, stereo |
| Integrated loudness | −13.91 LUFS | Spotify target: −14 |
| True peak | −3.55 dBFS | comfortable peak headroom |
| Loudness range (LRA) | 6.60 LU | soundtrack dynamics, not rock |
Correction on record: the internal sheet of 25/07/2026 noted LRA 6.8. Today's measurement returns 6.60. The other two values match the original note (−13.9 and −3.5). The measured figure stands, and the discrepancy is stated rather than quietly dropped.
Checksum of the measured file,
md5 = e4704e40d9831b52ddc9190ffb3ba75b. It is the same file delivered
to the distributor: between what was measured here and what is on the platforms
there was no mixing and no mastering stage.
The piece was written in honour of the artist's sister. That is the only thing this page says about it, and deliberately so: the rest is not public business. It is on record because it explains what the track is, and because a catalogue that publishes its method can also say, once, who a piece was written for.
The piece was going to be called The Force. It did not reach release under that name, and the reason has nothing to do with music: The Force is a third party's registered trademark in entertainment. Releasing an original composition under the one name capable of attracting a trademark discussion is paying dearly for a single word.
With the title changed, one question remained that a site promising method cannot leave open: the provisional name pointed at a known repertoire. Does the music point there too? That question, and only that one, is what the measurement below answers.
The title was a trademark matter. The melody was a measurement matter. Different problems, handled as two.
The Unbroken says what the piece holds from start to finish, and says who it is dedicated to. It keeps the "The ___" cadence that motivated the old name and owes nothing to anyone. The variant containing the trademark inside the title was discarded on purpose: inserting words does not neutralise a trademark, because the test is confusion, not literal identity.
None of the three checks was done by ear, and the first two use as references
recordings from the artist's own archive plus classical and
soundtrack controls, so that the result had something to be compared against.
Environment: basic-pitch 0.4.0 and librosa 1.0.0.
Automatic transcription, top-line extraction by skyline, conversion to an interval sequence (which makes the comparison immune to transposition) and a search for the longest identical contiguous subsequence. This is the method that catches the classic case: two identical bars in the middle of a different piece.
Cosine similarity between pitch-class profiles, 6 s window. Here is the part worth more than the result: without correcting for key, the references appeared to win comfortably (86.2% of frames). With key corrected, the advantage fell to +0.005 and 55.9% of frames, which is a coin toss. Transposed to the right key, Beethoven's Fifth scores 0.824, inside the same band as the references. All the apparent resemblance was E minor.
This is the kind of finding that only surfaces when the control is measured alongside. The first version of the number looked strong and was wrong.
The only one of the three that does not depend on a calculation of ours: it runs the real audio against the database of the real rights holders. Unlisted video, audio verified before upload (0.999964 correlation with the master, to make sure the encoding had not altered the fingerprint).
| Instrument | Result | Reading |
|---|---|---|
| Interval n-grams | longest identical run: 5 intervals (6 notes) | two-note oscillating figure, generic accompaniment material |
| Chroma, no invariance | references win in 86.2% of frames | misleading: the confounder is key |
| Chroma, with invariance | mean advantage +0.005, 55.9% of frames | indistinguishable from chance |
| Content ID | 1 claim, and it is the artist's own distribution | scan demonstrably executed; zero third-party holders |
The Content ID result deserves an extra line, because it is counter-intuitive: a video with no claim at all would have been a worse result. A blank result is ambiguous, it may only mean the scan has not run yet. Here exactly one claim appeared, issued by the artist's own distribution chain, covering 0:00 to 2:24, the entire track. That proves the system compared the complete audio against the database. Overlapping claims from different holders are routine in Content ID: had any segment matched a third party's asset, it would have appeared as a second row on the same screen. None did.
What goes online is the material belonging to the track itself:
And here comes the caveat a method page is obliged to give, even when it is uncomfortable. Three things were deliberately left out, and each one costs reproducibility:
Honest consequence of the three: this page is not reproducible end to end by a third party. It is auditable in what it publishes, the track's MIDI and the rule described above, and not in the rest. Anyone with a legitimate interest in checking the full comparison can request the complete package through the site's contact.
A report that does not declare its own limits is not a report, it is advertising. These are this measurement's, and the third is the most important:
GAP: it was not confirmed that any of them contains specifically the
piece the old title evoked. If the starting material were that one and no
reference contained it, those two tests would not see it. That is precisely why
Content ID was run: it does not depend on our references.Summing up: the audio does not support a derivative work, by three different routes, and none of the three proves absolute originality. That is what the measurement gives. The rest would be assertion without measure, which is exactly what this house does not publish.
The tritone in Insane Baroque Blaze: the report with an open detector →
Insane Baroque Blaze: track page →
Nando Valle: artist page →