Melodic Transcriptions

For every song in the RS 200, we have created a text file that contains a transcription of the entire vocal melody for the song. Our melodic transcriptions encode the melody using scale degrees, which show the relationship of each melodic note to the local key center. We tracked only the onsets of melodic notes, without any direct indication of the note length. In most cases, this length can be inferred from the onset of the next note. We did not use a recursive system to encode the melodies, since various iterations of the same section in a song often included a small (if not large) amount of variation. Our melodic transcriptions use spaces and line breaks to help visually format the files, but no real information as to phrase structure, section divisions, or formal areas is contained within our melodic transcriptions.

Since the melodic transcription task proved to take more time than the harmonic analysis task, we equally divided the melodic transcription work. We each (TdC and DT) transcribed the melodies for half of the RS 200, i.e., 100 songs each. (The songs were randomly apportioned between us.) We transcribed the melodies by ear, without consulting each other or any printed sources (e.g., lead sheets). In most situations, it was easy to determine which vocal part was carrying the melody of the song. Passages with multiple vocal lines, however, sometimes posed problems, in that it was not clear which vocal part was the melody. (One could certainly argue in some instances that both parts carry the melodic content.) In all cases, we forced ourselves to choose a single monophonic line to represent the melody. There was no particular methodology in choosing which vocal part would represent the monophonic melody, but we feel our decisions are at least one viable way of reducing the melodic content of a song to a single line.

To check for errors, we first ensured that our melodic transcriptions contained the same number of measures as our harmonic analyses. We also ensured that each melodic transcription was in agreement with the corresponding harmonic analysis file (by the same author) in terms of the key structure (main key and location of modulations) and meter (time signature and barline locations) for all points in the song. We then generated MIDI files for each song from a merger of our harmonic analysis and the associated melodic transcription (with the chordal root in the lower voice and the melody in the upper voice). Listening to these MIDI realizations, we could identify and then correct any typos or other errors in our melodic transcription files. After numerous rounds of revisions, we feel confident that our transcriptions reflect the melody as we are currently best able to transcribe it.

Our melodic transcriptions of the RS 200 can be downloaded here in a few formats. (The original author is indicated via a "dt" or "tdc" label at the end of the filename.)

The timed note list files were created by running our melodic transcriptions through the and scripts. Explanations and the original code for these programs can be found on our programs page. Note that each melodic transcription should be referenced only with the corresponding harmonic analysis by the same author. (This will ensure that the scale degree notations are interpreted correctly, since we sometimes had differences of opinion as to the location of key changes.)

The Melodic Transcription Notational System

  1. Basic Syntax
  2. Formatting
  3. Special Symbols

A. Basic Syntax

Our melodic transcriptions use scale degrees to represent pitches within a key. An integer indicates the specific degree of the major scale, and notes outside the major scale are notated via the addition of sharps ("#") or flats ("b").

At the beginning of every melodic transcription, the key is notated using the same bracket system as in our harmonic analyses, e.g., "[C]" indicates a pitch center of C. (We take "key" to simply mean a pitch center, without any particular indication of major or minor.) We also allow what we call "mode signatures." A mode signature indicates which of the seven diatonic degrees are always to be treated as flattened (in relation to major mode). For example, [......b] indicates that just the 7th is flattened (Mixolydian mode); [] flattens the 3rd, 6th, and 7th (Aeolian mode). If no mode signature is stated, major mode is assumed. Only degrees 2, 3, 5, 6, and 7 may be flattened (in any combination) and only flat symbols are allowed, not sharps (allowing other possibilities did not seem useful in rock). Mode signatures can be inserted at any point, just like key symbols. When a new key symbol is declared, the mode reverts back to major. (DT uses mode signatures in his transcriptions, TdC does not.) The n symbol can be used to raise a flattened note, e.g. "n7".

Like our harmonic analyses, we indicate bar lines with the pipe symbol ("|"), which indicates the end of a measure. (The first bar of a song thus does not require an initial pipe symbol.) At the beginning of every melodic transcription, the meter can be notated using the same bracket system as in our harmonic analyses, e.g., "[12/8]" indicates a compound quadruple meter. (If no meter indication is present, we assume a meter of 4/4.) The rhythmic/metric location of pitches within the measure is determined by the rule that each measure is equally divided in time by its contents.

Our melodic transcriptions only indicate the onset of pitches. To a certain extent, note offsets can be inferred from the fact that a note must have ended by the time the next note begins. But more subtle aspects of note lengths are not represented in our melodic notation. We indicate the lack of a note onset via the dot symbol, ".", which allows us to notate anything from simple to complex rhythmic divisions of a measure. A measure with no note onset may be indicated by a single dot and a pipe, or simply by an additional pipe.

As a simple illustration, the first four bars of "Ode to Joy" theme are transcribed below:

[D] [4/4] 3345 | 5432 | 1123 | 3..22... |

Our melodic transcriptions also contain information as to pitch height and melodic contour. At the beginning of every transcription, the register of the first melodic pitch is notated via a bracket symbol that corresponds to the octave in scientific pitch notation. Each following scale degree is assumed to be the closest pitch representative of the scale degree to the previous pitch. (Tritones are assumed to be ascending.) A leap to a pitch an octave above or below the closest representative can be indicated with "^" or "v", respectively.

For example, the following hypothetical song opening correponds in the first measure to an ascending scalar passage beginning on middle C (C4), followed in the next measure by a descending scalar passage beginning on the C an octave above middle C.

[C] [OCT=4] 1234 | ^1765 |

Changes in key can be notated by the insertion of a new bracket statement at any point within the measure. New time signature statements can only occur at the beginning of a measure.

B. Formatting

Unlike our harmonic analyses, we do not employ a recursive system in the melodic transcriptions. Every bar of the song is thus directly represented in our notation. This process can create long and dense files. Accordingly, we allow line breaks within the file in order to visually organize the transcription. A single line break typically corresponds to the end of a phrase, whereas two successive line breaks typically represent a larger division, such as a new formal section. That being said, we did not adhere to any particular methodology when inserting the line breaks; they were inserted simply for ease of reading and should not be taken as any particular analysis of the form, phrase structure, or hypermetric structure of a song.

Line breaks may occur at any point in the melodic transcription, as long as the line break does not occur within a measure. The following notation is thus not valid, since the first line lacks a pipe symbol, meaning that the third measure is split between the first and second lines.

[A] [OCT=4] 5.5. | .55. | 5.
5. | .55. | 5.5. | .55. | 5.5. | ..5.5..3 | 

5.5. | .55. | 5.5. | .55. | 
5.5. | .55. | 5.5. | ..5.5..3 | 

Other than the ill-placed line break within the third measure, the notation above is valid. Note that the double line break divides the transcription into two sections. The second section contains no errors in line break usage, with a single line break dividing the eight bars into what appears to be two four-bar phrases.

We also allow spaces to be inserted at any point within a measure. We use spaces informally to indicate the division of a measure into beats. (Spaces are not required, but they help to visually parse the transcription when reading and editing the encoding.) In measures with only basic divisions, we generally avoid using spaces. But as divisions reach low-levels of the meter (e.g., 32nd notes), we find that spaces to aid immensely in readability. The following standalone measures are all valid:

[4/4] 34343434 |
[4/4] 34 34 34 34 |
[4/4] 3... .34. 4.3. 3... |
[6/8] 3.. 4.. |

When parsing the melodic transcriptions for the sake of statistical calculations, all spaces and line breaks are ignored. Some conventions (such as the rule against line breaks within a measure) are enforced simply to guard against annotation errors. Note that pipe symbols, bracket symbols, and multi-rests (discussed below) all require spaces before and after unless these symbols are at the beginning or end of a line.

C. Special Symbols

A few symbols have special meanings.

Some special symbols are inherited with the same meaning from our harmonic analysis system. In particular, the "[0]" symbol indicates a lack of any clear meter. Also, the "%" means that everything afterwards on that line is a comment.

We also introduced the symbol "R" for rests in our melodic transcriptions, even though note offsets are not directly indicated. This rest symbol comes is useful when there is a long span of measures without any melodic content. In a similar yet different method to our harmonic analyses, we notate multiple measures of rest via the rest symbol "R", an asterix, and then the number of measures of rest. This multi-rest symbol requires a end-of-measure pipe to be valid. The example below represents a bar of melodic content, followed by 34 measures of rest, followed by another bar of melodic content:

[D] 3653 ...4 | R*34 | ..5. 33.7 |

Note that a single empty measure can be notated in a number of different ways, e.g.: " R | ", " R*1 | ", " . | ", or just " | ".

We also introduced a set of symbols to account for long strings of dots in a measure. Certain songs, particularly those with slow tempos, tended to have rather free vocal rhythms or vocal rhythms that were syncopated at low subdivisions of the measure. To increase readability, we adopted the following "multi-dot" symbols:

The dash symbol ("-") is equivalent to three dots.
The underbar symbol ("_") is equivalent to four dots.
The equal sign ("=") is equivalent to six dots.

Note that the actual rhythmic value of a multi-dot symbol varies (like the dot symbol itself), based on context. Consider the following isolated measures:

[4/4] _ 2345 |

(Here, the underbar is equivalent to half the measure, i.e. a half note in traditional notation.)

[6/8] 34_ 3.7.3. | 

(Here, the underbar is equivalent to a quarter note in terms of traditional notation.)

To avoid confusion, we generally used the multi-dot symbols within beats. (In fact, TdC used the multi-dot symbols only when they could account for an entire beat.) The following two bars are identical in terms of meaning; the only difference is in the notation, where the long string of 18 dots (equivalent to a beat in 6/8) in the upper measure has been replaced by three equal signs in the lower measure.

[6/8] 3.23.23....21..... .................. |
[6/8] 3.23.23....21..... === |