Frequently Asked Questions (FAQ)
Click here to read a press release about the Melody Comparison and Rhythm Comparison "copyright checker" tools.
How do I use Dr. Lustig’s Rhythm Comparison tool?
First, you choose the tempo/BPM of your rhythm. You need your sound on to hear the metronome, which is critical for inputting your rhythm. Then you click Record and tap your rhythm along with the metronome, clicking Stop when you are done.
Why can’t I speak the rhythm or upload the audio?
The current technology for accurately converting audio inputs to meaningful music-symbolic data just isn’t good enough yet to get reliable results. For that reason, for the Rhythm Comparison tool you can tap the rhythm. Instead of an audio copyright checker or mp3 copyright checker.
What is a problematic similarity percentage between two rhythms? How much is TOO much?
There is no easy answer to this question because it depends entirely on the case and context. For example, two songs’ rhythms might both go quarter-quarter-quarter at one point—which is technically a 100% rhythmic similarity—but if those three notes are the only commonality between the two songs, and they occur with different pitch series, the similarity isn’t particularly significant. Further complicating the case is the idea of “prior art”: the musicologist can easily demonstrate that the pattern quarter-quarter-quarter has occurred in a lot of prior music that predates both the plaintiff and defendant, and therefore is not protectable in the first place.
On the other hand, sometimes two songs can be in different genres and sound quite different, and yet expert forensic musicology analysis can shine a light on the hidden similarity, helping to reveal and demonstrate that highly distinct/original rhythmic aspects have been copied from the plaintiff’s song.
How do you compare two rhythms?
The algorithm finds the shortest note value in the inputted rhythms, and subdivides the rhythms to this resolution, creating a binary vector of attacks (note onsets) or non-attacks (i.e. rests or sustains: no distinction between the two). The two binary vectors are then compared to determine their overall similarity percentage based on how many points in the vector share attacks. Unlike the Melody Comparison tool, Levenshtein distance is not leveraged - because a straight rhythm (e.g. X. X. X. X. - every note landing on the beat) vs. a syncopated rhythm (e.g. .X .X .X .X - notes landing off of the beat) are entirely different things perceptually.
Why do I need to use the metronome for the Rhythm Comparison tool?
The metronome is necessary to know what your tempo is and where your beat lands. Otherwise, your taps are impossible to decipher. Rhythm depends on having a beat/pulse and tempo. To compare two rhythms, the algorithm needs to know what rhythmic durations are involved (e.g. quarter note, eighth note).
Why are the rhythms notated as individual note durations instead of beamed together, showing the beat structure, etc.? Why are there no barlines? No rests?
The user taps the rhythm along with the metronome. Placing barlines requires knowledge of where “beat 1” is. While many rhythms do not enter right on beat 1, the average non-expert musician may not be aware that their rhythm doesn’t come in right on beat 1. (Also, the system would have required an additional level of complexity: that the user first specify beat 1, and then proceed to tap.) Instead, therefore, the algorithm actually has no concept of beats or bars - it simply compares two rhythmic series. In line with most theories of musical rhythm, the algorithm considers only onsets, not offsets - that is, it doesn’t care about the durations of notes. It only cares about where new notes arrive. There are no rests because it’s impossible to know only from input taps whether they represent long notes, vs. short notes followed by rests. Besides, the general perception of rhythm usually comes from the note onset profile.
Click here to read a press release about the Melody Comparison and Rhythm Comparison "copyright checker" tools.
How do I use Dr. Lustig’s Rhythm Comparison tool?
First, you choose the tempo/BPM of your rhythm. You need your sound on to hear the metronome, which is critical for inputting your rhythm. Then you click Record and tap your rhythm along with the metronome, clicking Stop when you are done.
Why can’t I speak the rhythm or upload the audio?
The current technology for accurately converting audio inputs to meaningful music-symbolic data just isn’t good enough yet to get reliable results. For that reason, for the Rhythm Comparison tool you can tap the rhythm. Instead of an audio copyright checker or mp3 copyright checker.
What is a problematic similarity percentage between two rhythms? How much is TOO much?
There is no easy answer to this question because it depends entirely on the case and context. For example, two songs’ rhythms might both go quarter-quarter-quarter at one point—which is technically a 100% rhythmic similarity—but if those three notes are the only commonality between the two songs, and they occur with different pitch series, the similarity isn’t particularly significant. Further complicating the case is the idea of “prior art”: the musicologist can easily demonstrate that the pattern quarter-quarter-quarter has occurred in a lot of prior music that predates both the plaintiff and defendant, and therefore is not protectable in the first place.
On the other hand, sometimes two songs can be in different genres and sound quite different, and yet expert forensic musicology analysis can shine a light on the hidden similarity, helping to reveal and demonstrate that highly distinct/original rhythmic aspects have been copied from the plaintiff’s song.
How do you compare two rhythms?
The algorithm finds the shortest note value in the inputted rhythms, and subdivides the rhythms to this resolution, creating a binary vector of attacks (note onsets) or non-attacks (i.e. rests or sustains: no distinction between the two). The two binary vectors are then compared to determine their overall similarity percentage based on how many points in the vector share attacks. Unlike the Melody Comparison tool, Levenshtein distance is not leveraged - because a straight rhythm (e.g. X. X. X. X. - every note landing on the beat) vs. a syncopated rhythm (e.g. .X .X .X .X - notes landing off of the beat) are entirely different things perceptually.
Why do I need to use the metronome for the Rhythm Comparison tool?
The metronome is necessary to know what your tempo is and where your beat lands. Otherwise, your taps are impossible to decipher. Rhythm depends on having a beat/pulse and tempo. To compare two rhythms, the algorithm needs to know what rhythmic durations are involved (e.g. quarter note, eighth note).
Why are the rhythms notated as individual note durations instead of beamed together, showing the beat structure, etc.? Why are there no barlines? No rests?
The user taps the rhythm along with the metronome. Placing barlines requires knowledge of where “beat 1” is. While many rhythms do not enter right on beat 1, the average non-expert musician may not be aware that their rhythm doesn’t come in right on beat 1. (Also, the system would have required an additional level of complexity: that the user first specify beat 1, and then proceed to tap.) Instead, therefore, the algorithm actually has no concept of beats or bars - it simply compares two rhythmic series. In line with most theories of musical rhythm, the algorithm considers only onsets, not offsets - that is, it doesn’t care about the durations of notes. It only cares about where new notes arrive. There are no rests because it’s impossible to know only from input taps whether they represent long notes, vs. short notes followed by rests. Besides, the general perception of rhythm usually comes from the note onset profile.
Contact Dr. Lustig below:
Disclaimer: the results of this tool do not constitute official legal or musicological advice. Data from this page such as the “percent similarity” metric cannot be used in legal documents and does not constitute an endorsed statement by Dr. Ethan Lustig. These tools are previews and not official analyses of similarity. For a real expert interpretation, contact Dr. Ethan Lustig.
