The Voynich Ninja

If you write out the words and melody of a song, breaking the lyrics into syllables, the resulting "words" have unusual structure. A simple example to explain what I mean:

twink1 le1 twink5 le5 lit6 tle6 star5long how4 I4 won3 der3 what2 you2 are1long

The numbers are the degrees of the scale, which is not, I believe, how medieval music theory worked, but this is just an illustration. Another example:

ut1 que2 ant4 la23 xis2long re2 so2 na1 re2 fi3long bris3long mi35 ra3 ge2 sto31 rum2long

For this latter example, I used a real historical song in Latin (albeit one significantly earlier than the Voynich MS), which contains more melismas (syllables extended over multiple notes).

I arranged the consonants in a way that seems more natural to me as an amateur singer: in the places where they are pronounced. So, fi-bris, not fib-ris. This could cause syllables to have more internal structure. A language might normally allow many consonants at the end of syllables, but in this writing method, most final consonants are only allowed to stay in that position when there is no following syllable to attach them to instead.

Music notation has items which appear at the beginning of sections and rarely elsewhere: clefs, section names, part indications in polyphonic music. Also, there is a symbol that appears at the end of lines in some historical notation, the You are not allowed to view links. Register or Login to view., which even vaguely resembles the EVA "m".

I believe that melodic markings on text are historically known, such as You are not allowed to view links. Register or Login to view. .

You are not allowed to view links. Register or Login to view. mentions that there have been attempts at a musical interpretation - can you point me towards any further information?

I think the fact no one can read the text gives the impression it is chaotic, the issue is that it is too predictable and the alphabet is so short there is definitely no room for numbers and a fully working alphabet. Anyway, the music of Hildegard Von Bingen might be of some interest.

To rephrase bluetoes' point: this idea presupposes that the VM text consists of normal syllables that have been altered by the addition of repetitive parts. But there are no normal syllables in there. If you set aside enough common suffixes to write music, you'll still be left with a poor text. (It would be worth the experiment though).

I doubt whether the method of splitting syllables has much of an impact. Fi sounds better than fib, sure, but ris sounds better than bris. You're just pushing the complexity into the other syllable.

Regarding music, there are many ways to go about it, but I've never seen anything even remotely convincing. I can't really recall a serious attempt at a solution this way. There was one guy with a YouTube channel who believes that it's music, but his method was really dumb: use simple substitution to turn it into sound, then set the result to a melody you like and hire an unfortunate choir to sing it.

(03-05-2025, 11:58 PM)extent_of_foxes Wrote: You are not allowed to view links. Register or Login to view.there have been attempts at a musical interpretation - can you point me towards any further information?

You are not allowed to view links. Register or Login to view.
You are not allowed to view links. Register or Login to view.

Thank you for the links!

Good point that there isn't room for a full alphabet in addition to musical notation. I can come up with some ideas to address that, but I want to be clear that my main point was just the observation that this is a way to generate "words" which have unusual structure, not any sort of specific detailed theory. To carry on with wild guessing:

• There might not be lyrics at all, if the music doesn't have lyrics, or if the music is intended to go with a text that the reader is supposed to be familiar with.
  
• There might be only some sort of reduced lyrics, to remind the singer how the music and text line up.
  
• The lyrics might not include all phonetic distinctions present in the language. For example, voicing distinction might be dropped, or sounds that the author disliked might be removed entirely. This would be a odd thing to do, of course.
  
• And, to reduce the number of symbols needed for the music, the pitches or rhythm might not be precisely specified.
  

Regarding "Fi sounds better than fib, sure, but ris sounds better than bris", I'm not actually talking about what sounds better, but about how a singer might consider syllables to be divided. Consonants must be sung at some particular time, or the choir will sound disorganized. As I personally think of it (I have some early music experience but I'm fundamentally raised in a modern choral tradition) the word "fibris" does not have two notes. It has six! The initial unvoiced "f" is a grace note before the first sung pitch, "i". The "b" and "r" are a pair of grace notes before the second held "i", and if the two syllables were sung on different pitches, the "b" and "r" would have the pitch of the second syllable. Finally, the "s" has to be placed somewhere, either on the next syllable, if there is one, or at the end of the last note of the phrase. A consonant at the beginning of a syllable can be timed in reference to the held note, but a consonant "floating" at the end requires (for me) keeping track of a separate rhythmic event.

Ah yes, I see. Reminds me of the Gloria meme:

(04-05-2025, 08:47 AM)Koen G Wrote: You are not allowed to view links. Register or Login to view.There was one guy with a YouTube channel who believes that it's music, but his method was really dumb: use simple substitution to turn it into sound, then set the result to a melody you like and hire an unfortunate choir to sing it.

This guy by any chance?
You are not allowed to view links. Register or Login to view.

Yeah...

The Gloria meme is pretty funny.

It gives me a serious idea, too, to add to the list: maybe there are lyrics from a real language, but the lyrics are extremely simple or repetitive. If there are only half a dozen distinct syllables in the lyrics used in the entire MS, then the author might have used abbreviations for them... which might mean no chance of figuring out the plaintext.

Even if we can't tell what the words are, there could still be some statistical properties we could use to test this idea:

Given a proposed division of the text into "lyrics" and "non-lyrics" parts, it should be possible to combine most lyrics-parts with most non-lyrics-parts and get valid segments of Voynichese. Like if a whole book was filled various settings of "twinkle twinkle little star", the syllable "twi" should eventually appear with most pitches and rhythms.


Given some way of extracting just the lyrics parts, they should be made up of a sequence of chunks which almost always appear in the same order, optionally with some sub-sequences repeated, or repetitions of the whole sequence. And some sub-sequences should be much more common to repeat than others. For example, if the plaintext lyrics were "gloria in excelsis deo", then "gloria in excelsis deo gloria in excelsis deo" should be very common, and "gloria gloria" should be common enough, and "glo- gloria" might be possible but should be pretty rare, and "excelcelcelsis" should not appear at all unless the scribe made a mistake.

If the rhythm is notated precisely, which is definitely not a reliable assumption, then there might be sections of the text, physically near each other, which add up to the same total duration. (Verses or polyphonic parts.)

I wonder what the distribution of various pitch and rhythmic indicators is in known music notation systems...

Thanks Koen, I shouldn't reply to stuff at 2am  . 
Oddly the guy you mentioned is also the one I had in mind when saying "music of Hildegard Von Bingen".

If I were to try and make music from the text and avoid all the pitfalls of a system I would do something like this. 
I have a feeling that it may not be very musical.. but it would make noises and have words

[attachment=10540]