Introduction


The Focus and Objectives of these Pages

The Internet is huge and and the MIDI world is large, but the world of Classical Music and Midi on the Internet is merely of respectable size. Karaoke on the Internet is not vast, while the combination of Classical Music, Midi, and Karaoke is downright minuscule. In fact this site, which is still very much in the development stage, might very well already represent more than half of such a world. Its alliance with its older companion site, Emily Ezust's Lied and Song Texts Page, which contains a vast collection of texts to art songs of the world, lends it a kind of archival body and pedigree, especially as Ms. Ezust is the No. 2 developer on this site as well as the director of her own site. Both of us want to do what we can to promote the universe of the Art Song, which we extend to include all music that might be considered essentially "classical" and that involves words.

It is because of the words that our site is offering one aspect of a musical experience that is actually superior to that found in the "real" world, namely the synchronization of the meaning of the sung music to the moment of its reception. Thirty years ago, I (R.E.C.) wrote a ten-page treatise on how and why live Opera performances should have visible texts presented, and sent 100 copies out to the members of the National Opera Association. There was no real reaction. Around 1983, such a movement sprang into existence with immediate audience approval. This movement has not, however, been extended as yet to the stage recital; therefore, where every other aspect of such a recital is decidedly superior to listening to records or to MIDI, unless the singer has the very finest diction and is singing in a language that the audience understands, the foundation of the song - the text - is, on the receiving end, mushy at best and opaque at worst. We try to encourage a synchronization of this understanding.

Fundamental Purpose

The musical and computer worlds meet in MIDI and form a kind of spectrum. Unlike the spectrum of our political electorate, with the bulge in the middle, the above connection is attenuated in the middle. In other words, the techie types at one end and the purist classicists at the other have little do do with one another. Our page seeks to redress this gap and embrace both, but not out of a love of tolerance. On the contrary, I feel stongly that the modern aesthic that proclaims anything is Art (while subject to specific valuations), is nothing more than a repudiation of aesthetics itself. Therefore, on this site we will not tolerate just anything, although we mean to be pretty comprehensive vis-a-vis Classical Music in its most broad definition. By this we mean to exclude all forms of what would generally be recognized as the commercialized popular music of the last 35 years. The fundamental objective is to refamiliarize the Internet world with what is generally referred to as "Classical Music", especially Song, and to promote the development of computer tools for understanding, dissecting, and eventually giving rise to new creations based upon such tradition.

The Quality of MIDI is Sometimes Strained

MIDI files now available on the Internet, including those at this site, vary considerably in quality. None of them is as good as the real thing, and in this comparison we will consider live recordings as a subset of "the real thing". So what's the point? The point is that with the right equipment and software you can participate. The only advantage to the passivity of listening to MIDI files is that you can perhaps access a range of material that you couldn't easily obtain otherwise, plus, on this site, the situation of the word presentation discussed above. If the examples are really crude, however, it is likely that the the loss outweighs the gain. So we must start by recognizing that we are gaining and losing at the same time. It should be noted that music has always been in this situation. Most instruments only do some things well, not everything. About the only perfect instuments are the voice (sometimes) and bowed strings, and the latter contain the problem that they are weak volume-wise in a large ensemble and it is only because they are so fine that the expense of their numbers is tolerated.

The bad news:

    1. We cannot offer .wav files (full recordings) because they take up too much space (and we are thinking in terms of massive archives). We are open to discussion here from any of our client users.
    2. This archive is for general access, but only a minority of the Internet world has the sound equipment to listen to anything with decent quality.
      This is likely to change in the next few years. At the moment, even the Sound Blaster Card is not good enough, at least in its default configuration. We would rather not offer only pieces for advanced sound modules that are not widely owned, and we would rather not use sound banks that are not somewhat standard. We will aim, for now, at the Roland Sound Canvass which, if not high end, is decent. We will stick to the General Midi sound bank, following the lead of "The Classical Midi Archives", which is the leading site for Classical Music and has established a kind of norm in this genre.
    3. We feature vocal music, but we have no actual voices.
      That is, we have to use clarinets, oboes, flutes, etc. for the vocal patch track. If our site springs to life with good input from elsewhere, we would welcome some guidance on the use of sampled real voices to be used as dumps to the sound module (for use as ordinary MIDI patches). They would have to be effective with Roland-Soundblaster equipment, however.
    4. Our staff notation is mediocre at best. Our emphasis is on words, not notation; nevertheless, we would like to have the vocal line look right, so that a singer could practice with our material and alternately listen to and mute the vocal track while watching the notes (the rest of the score is not to be troubled with for now). We are only dealing with MIDI files, so this staff must generate from a MIDI file. We generate our pieces through Cakewalk, but Cakewalk refuses to use flags and beams based on syllables as per vocal convention. It also gives no grace notes. I believe both of these can easily be overcome in software, but I am not aware of their availability yet.
    5. We are short on real time offerings.
      At the end of 1996, we have still not enlisted pianists familiar with the Lieder repetoire to work with us regularly. We have a few such in the Schubert section of our site. I believe this is the way to go for great hunks of the Lieder repetoire. Not only will the piano be distinctly superior to anything that can be done in step-time, but the work on tweaking and shaping should be minimal, and that only involving the vocal track on occassion. If the vocal staff is to be presented well, it will have to be entered in step-time, and the lyrics windows will have to be presented entirely line by line, not Karaoke style. This all has some priority.


MIDI Pieces in Step-Time

The majority of pieces in raw step-time are insufficiently musical and some are downright terrible. You can get away with Bach; even Beethoven does not sound too bad if the instrumentation is well balanced, but in general the dead regular beat and clockwork-even velocities need to be disguised somewhat. It is easy to change the tempii in Cakewalk, and a crude approach, better than nothing, is just to paint a shallow wavy line across the straight tempo line. This is, of course, quite random. Better yet, the tempi should be followed, either from the score, or from memory, bit by bit, when the creator knows the piece well. Sometimes, even with a score, such is necessary since interpretations are not always indicated in the score. Then there are velocities. Ideally, each note would have its own velocity as per a real-time performance. A compromise result can be achieved by "painting" the result in the piano-roll window in Cakewalk. Accents must be done individually. Also, in Cakewalk it is easy to lengthen or shorten a block of notes. This has the effect of creating a staccato or legato effect for that group.

Full scores require a lot of work. A compromise is to use a piano score and perhaps make duplicate tracks of horizontal parts and then assign various patches at various times to these tracks. All of the above requires artistic taste. Such interpretations should be copywriteable, where simple step time notation may not be. More sophistication can be achieved by dealing with the controllers to modify certain notes. The Wesendonck group and some of the Mahler have been dealt with in this manner; Mahler is difficult to do, but we have some offerings; and we are currently working on some Berlioz as well. These are all art songs for orchestra that are sometimes played on piano. The above approach is a compromise to a full score, but in fact it often sounds better even than a real-time presentation on piano.

All of this tweaking is tedious, but can be fun in a way, and quite instructive. In the future, such results will need to derive from real-time sensors, the same way velocities do now. It is only because the piano does not need a whole lot more than velocity to be respectable that it is here now for real time archives. Almost everything else requires tweaking. The biggest drawback of all in the Synthesis world, especially applicable to vocal music, is the lack of a kind of proper pitch slide that singers use all the time. Forget the modulation controller. No good. This needs also to become real time, but I haven't seen it, even artificially, used on a note by note basis.

The theoretical plan for this site, when we develope a real community, will be to encourage a constant upgrading of the pieces, an upgrading which would be carried in several stages. This will require many authors, all of whom will be given credit, but who would likely not be able to own the copyright to their efforts. In addition, we might be able to work out a way they could participate if any money were generated. Contributors who do want to own the copyright will have to do most of this work themselves. This site encourages contributions, but will not be looking for complex pieces unless they are either properly enhanced or made available for enhancement. Pieces that are fairly simple and robust, such as the Burns songs offered only require a fairly easy enhancement, so one contributor could retain the copyright for his/her offering. On the foreign-language texts, the translator would own the copyright to that translation.

In conclusion, our site hopes to be ever-expanding and ever-progressing with the diverse help of the Internet community. Please feel free to experiment with our files and submit your own interpretations for us to put up!


Back to the Classical Midi with Words Page
Robert E. Crawford, recrawfo@recmusic.org