WordBuilder: converting lyrics to real music

[AndyFinkenstadt]

A short recipe to match a technique I'm still refining:

Let us suppose you want to take a song that has English lyrics, such as SILENT NIGHT. If there were always one note per syllable and one syllable per word , this would be rather easy. But neither English nor life is like that. So, we need to do some doctoring.

My setup for this is: radium 49 keyboard for input, word builder listening on the USB KEYSTATION port and sending to Maple MIDI 1, and Choir's kompakt listening on Maple MIDI 1 and not outputting anywhere. Both word builder and the choirs are set for boys, using the "F BOYS WB HARD" multi-instrument.

First, key in the lyrics for silent night:

Silent Night Holy Night
All is calm All is bright
Round Yon Virgin Mother and Child
Holy Infant So tender and mild
Sleep in heavenly peace
Sleep in heavenly peace

Next, using the keyboard, just play 1 note per word, listening for pronunciation, or if you are familiar with Votox, just inspect its output. Fix any typographical errors.

Now, switch to VOTOX mode by clicking on the "VOTOX" button beneath the "ENGLISH" button. This is the base result:

SaElunT naET HoOlEE naET
ol Ez Kam ol Ez braET
ruOnd yan vurjEn mut!ur And C!aEld
HoOlEE EnFunT SoO Tendur And maEld
SlEEP En HevunlEE PEES
SlEEP En HevunlEE PEES

Step one, run through in your mind the melody of Silent night. It goes Si-i-lent Night. The long-I sound happens to be a dipthong consisting of TWO vowel sounds: a (like in far) and E (like in sleep). When dipthongs are sung, the first vowel sound goes as long as possible, even across slurred notes (except with a special barbershop technique called showing off is done :)), so the VOTOX equivalent to Si-i-lent night is: "Sa aE lunt naET". Any time the melody changes notes while keeping the same vowel sound, duplicate it in the Votox. Anytime there is a change in syllable, split the word with a space. Thusly:

Sa aE lunT naET Ho oO lEE naET
ol Ez Kam ol Ez braET
ruOnd yan vu ur jEn mu t!ur And C!aEld
HoO lEE En FunT SoO Ten dur And maEld
SlEEP En He vun lEE PE EES
SlE EEP En He vun lEE PEES

(This is, at the moment, untested, since WB 1.0.3 is crashing on my machine while doing this tutorial. But it worked last night before the upgrade.)

Now, just play silent night on the keyboard. Use a smooth technique, but make sure you lift fully from the key prior to playing the next one. (Similarly, if you are using a sequencer, make sure that your notes end before the next ones begin. And if you are using Sibelius, set the performance settings for durations of slurred notes to 99%.)

[neoTypic]

Bravo. This is a great start and what I have found myself to be the best way of going about things so far.

I've been having trouble getting the legato / staccato attacks to work myself. I need to look at the instruction manual again. :p

[Thomas]

Very good, Andy - that's exactly what I have been doing as well. Pasting or typing in the words in English and then switching to the Votox window is a great way to start. Sometimes when I dont like the sound of a particular syllable, - because I don't know Votox too well - I go back to the English window and type in another word that rhymes with the word (or part of a word) that I want. Then I switch to Votox and eliminate the letters in the rhyming word that are not part of the real word. This may sound stupid, but when just learning Votox it is a quick way to get a sound I want. After a while I will probably make all adjustments in Votox.

Now, I am sure you know that when a syllable is sung over several notes the method you use (putting in an extra repeated vowel) works fine, but is not the only way to do it. The manual (p.77-8) describes a different way. I have tried this, and it works. It is a good program idea, but still needs work in the implementation. Here is what I did:

In Wordbuilder go to Options-Events and set "Hold Syllable On" to CC21 value 1, and set "Hold Syllable Off" to CC21 value 2. (You can use other values if you want - these are just the ones I use). Then hit Tools-Save Options.

Now you only have to type in each word once, and put in the CC codes in your sequencer at the start and end of any combination of notes that are to be sung on one syllable, (as notated by a slur in vocal music.) In Sibelius this could not be any easier. All you do is select the first note of a slurred combinaton and hit CTRL-T and then type in ~CC21,1. Select the last note of the slurred combination and Hit CTRL-T and type in ~CC21,2. (CC must be capitalized). Now Wordbuilder will automatically prolong the one syllable over those notes, no matter how many are in between.

This method should be better because you don't need a different set of words for each part (if parts slur at different times). It is not better yet, however, because WordBuilder does not know what to prolong and what not to. I have noticed that it works great only on pure vowels, but does not make the right choices with dipthongs, or complex words with beginning and end consonants. If you prolong the word "I" for example, it seems to be automatically play "aE aE aE aE", when it really should be playing "a a a aE". Maybe the definitions of this will be improved in future updates, because this feature has quite a lot of promise. Try it out as an experiment and let me know what you think. At the moment, writing in your own extra vowels sounds smoother and more natural, but this way of doing it would be better in the long run if the program could get the definitions a little better.

By the way, as long as you are using the control codes, I have found the command ~CC20,127 to reset the position in Wordbuilder to be great. That is, put this command on the very first note of your piece in Sibelius, in the same way as described above, and every time you start from that point wordbuilder will begin with the correct words, without you having ot hit "reset position" or F5 before starting every time. Saves a lot of time.

If you are using a sequencer or anything else instead of Sibelius, you can also put in the CC codes and do all this, - only the method of entereing them in your midi file will be different.