SongTrellis |
|
||||||||||
|
|||||||||||
New Changes since December 3rd:
New original compositions since Dec 3rd Bernard Chinn submitted a wonderful original to SongTrellis on December 15: Funtime for Flutes Joe Jagelka sent in three on January 12th that are great work: No. 5 Schoolhouse Lane, One Foe Sergio P. and Kathy My Love. On Monday he submitted two pieces composed using the Musinum fractal music generator: East and West. Folks, I've been remiss in someways and faithful in others. As you can see it was December 3rd, more than a month and a half ago when I last updated the SongTrellis home page, quite a long run of being remiss in that department. On the other hand I've worked on a number of things that may eventually be for the greater good. One thing you should notice immediately: the voicings of chord progressions that are generated by the SongTrellis editor and posted to the site should be musch more musically pleasant to listen to. You may not realize it, but 90% of the chord progression arrangements on this site have been produced algorithmically by the SongTrellis music editor without human intervention. We enter chord sequences by specifying the root name, type of a chord and duration for each chord in a sequence. By this method, it usually takes about five minutes to spell out all of the chords for a new progression. We don't even bother to spell out the individual pitches used in the chord. When we press the Play button on a score's window, the editor chooses pitches for the chords and lays them out in a way that should seem musically continuous, all in a fraction of a second. We usually replicate the chords for a tune 6 or 8 times, press Play and save the results as a MIDI file which we post to the site. We take the voicings used in the first repetition of the chords and use the editor to create a picture of those for the progression's score. Many times it takes less than 20 minutes from the time I start working on a chord sequence to the time it is posted onsite and indexed for The Changes listing. Over the Christmas holiday, I changed the editor's chord voicing alogorithm to use better chord voicing rules. If you look at the links in the calendar on the right side of the homepage, you'll hear that the new chord progressions submitted to The Changes starting Januray 4th somehow sound much better than those submitted before then. The algorithm I invented previously didn't pay enough attention to how the important tones in a voicing (usually the 3rd and 7th intervals) should be placed to lead your ear to the important tones used in the next chord. The old algorithm did its worst work when the tempo for a score was fast or when chords with more than 4 notes were used. In these situations the voicings seemed to change register too often and too unexpectedly. I hope your ears will confirm for you that the new method is a big improvement. The SongTrellis editor is now a Macintosh PowerPC app. Previously it was compiled to the 68000 instruction set, which is emulated on most Macs. This slowed the app down unnecessarily. Right after December 3rd, I converted all my editor code so that it could be built with the Metrowerks C compiler. I had put off doing this for a long time, mostly because Metrowerks' documentation was very sketchy and unhelpful. . The main reason I had to do the conversion was that Apple now ships all new Macs with their OS X operating system which requires that all application programs be compiled to Power PC code. For my code to have a future it needs to run on OS X. I am only part way to OS X compatability but doing this conversion was a large proportion of the work that is necessary to achieve this. The conversion took nearly three weeks but it lead to several benefits: the editor now is compiled to use the PowerPC instruction set so the app seems to execute many important operations (launch, menu drop, generation of chord voicings before score playback starts up) are twice as fast. Because I'm using the PowerPC instruction set, I can start changing my code so it will run on Mac OS X. I am also able to use a much better debugger so my pace of development is speeding up. I have a number of transcriptions of musical performances, which I like to study, that I've typed into SongTrellis Editor scores. When I press the Play button, the editor animates the scores to show which notes are playing as the music for the score is synthesized. Early in January I began to add features to the editor so that I can synchronize transcription animations to the actual performance of the solo recorded as an mp3 file. This is cool from a number of standpoints. When I'm actually writing down a transcription myself, I can check it by allowing the original recording and the MIDI performance of the transcription to play at the same time. Any place where the MIDI synthesis does not match, shows a place in the transcription where I have I written incorrect notation. Also, There are many subtleties of performance that cannot be encoded into a score. With synchronization you can hear how a performer actually plays a note as it is highlighted in the score, which helps you pick up many nuances of the performance. Finally, when a certain part of the score is selected and played you only hear the music in the performace that corresponds to the selection. This makes it much easier to focus in on interesting parts of the score. Even if you don't read music, you can follow the contour of a line recorded in the score and match it up with what you hear in the performance. My test case for MP3 synchronization is the solo Wayne Shorter played on the tune Crisis on Freddie Hubbard's album "Ready For Freddie". Elvin Jones is the drummer on this recording. Despite the fact he is laying down a complex polyrhythmic line his tempo is accurate to a thousandth of a beat per minute over the entire solo. By carefully choosing the moment in the performance that corresponds to the exact beginning of the score and choosing the correct tempo, I was able to get the score animation to sync closely with Wayne's performance. Since then I've used the feature to help figure out some of the chords used in Etta James' performance of At Last and to transcribe the melody of a very fast bebop line that Al Haig wrote called "In A Piinch". Please join our community at SongTrellis. Our contributors welcome your comments, suggestions and requests. As soon as you join the site (or login if you are a member) a response form will appear here.
|
|||||||||||
Last update: Friday, March 1, 2002 at 9:16 AM. |