Aug 13, 2011 Music
I have added an update to the song Jazzmine Revolution, because I felt it wasn’t quite right. I made some changes between 2:10 and 2:20. I don’t blame you if you don’t notice the difference at first, but the change gives the second part of the song a more natural flow.
You can tell apart the old version from the new version in that the old version’s file name is jazzmine.mp3 and the new version is jazzmine-revolution.mp3.
I’m working to get my new track on music streaming sites like Spotify and iTunes. If I can get my end of the process done next week, then it will be another 5 weeks for Spotify to be able to list my song on their site, so the song’s release date will be some 6 weeks from now, IF I hit no snags.
Update, Aug 23th: it seems I uploaded the wrong file. Forget about the file jazzmine-revolution.mp3, the correct version is JazzmineRevolution.mp3.
Sep 1, 2009 Music
I have used computers since 1980 and samples since 1990, but the most impressive piece of music related technology I’ve ever laid my hands on is Vocaloid. As a software instrument it’s still in the infancy of its development, because the text-to-speech aspect still leaves something to be desired in comparison with reality, so it cannot model a voice as realistically as a multi-sample can model a violin, for instance.
The true power of Vocaloid lies elsewhere, though. It begins with the concept of a multi-sample. In the old days a sound recording was taken of, for instance, a voice, which could be reproduced by pressing a key on a keyboard. If the singer had sang an A note, at say, 440Hz, and you would press the 440Hz A key on your keyboard, then the sound would be natural. If you pressed a key higher on the keyboard, then the sample would be transposed upward and the singer’s voice would sound if he were on helium. Multi-samples solved that problem, because recordings are made at every possible pitch, resulting in a package of short sound recordings that comprises the multi-sample. Then, a different sound recording (sample) is associated with each key on your keyboard. A multi-sample even consists of samples for different velocities, so if you press a key hard, you get a different sample than on a soft touch.
What this actually does is record the behaviour of an instrument or a singer. Each singer will produce a different set of multi-samples, even if their voices were identical, because each singer will respond differently to requests like: “Now sing a little louder, please”. I’m sure this is an uneducated, simplified interpretation of the actual technology, but I’m trying the set the stage for why I think the Vocaloid is so useful.
So, the Vocaloid takes the concept of recording a singer’s behaviour many steps further.
The data from the score file is sent to the synthesis engine, which draws on the phonetic and expression databases to create the track. To sing the word part, for example, the software combines four elements from the phonetic database: p (as it sounds at the beginning of a word), p-ar (the transition from p to ar), ar-t (the transition from ar to t), and t (as it sounds at the end of a word). The two ar elements are blended together, and the resulting vowel a is lengthened to accommodate the melodic line.
So the individual characteristics of the way the singer pronounces words is recorded, but also his style of singing and musical habits. This information is then stored in a database.
The mindblowing, fascinating aspect of Vocaloid is that this database is all-revealing. It says something about the singer’s cultural background, her musical background and even her personality. The thing with a Vocaloid is that it’s totally manipulable. You can make it sing and say anything you want, bringing out all the contents of the database. You can play with the musical style of the singer, how she reacts to various musical circumstances, revealing, as said, her musical background. Vocaloid Prima, for instance, was modelled after an opera singer and you can clearly hear the enormously rigorous and disciplined musical education she had. You can also hear that she’s a much more complicated person than Vocaloid Miriam (modelled after well-known singer Miriam Stockly).
Although Prima is an opera singer, she doesn’t have a big voice. It sounds a lot darker than Miriam and menacing, like Maria Callas. In classical music it has become fashionable to sound “jazzy”, with very clear, overtone-free soprano voices, turning a singer into a walking sinus wave. Prima, fortunately, has a very rich, complex voice.
For jazzy stuff Miriam is perfect and he has a very captivating, light sounding voice, very Northern, as opposed to Prima and Lola, the third female Vocaloid. Lola has a dark, clear voice, well-suited to pop music. I know that I will eventually buy all Vocaloids, because I’m hooked.
The thing with a Vocaloid is that as a music producer you get a very intimate relationship with the voice in question. The Vocaloid’s database contains such a wealth of information, that it’s Orwellian. The best way to immortalize youself is to have yourself Vocaloided, because the database will have your personality.
For me, as a musician, Vocaloid presents the opportunity to write songs, not just instrumental pieces, and to explore all the musical aspects of the singing voice in general and the individual singers in particular, all in a few mouse clicks. As opposed to a real singer, you can make a Vocaloid do anything, use her musical experience and explore musical pathways that are completely new.
The technology behind Vocaloid is Yamaha’s but the sound recordings of Prima, Miriam, Lola and Leon were done by Zero-G. They ask their customers for suggestions on what kind of voice they’d like to be modelled next, but I’d be more interested in: WHOSE voice? Because to me Vocaloid is about the individual musician, not about their kind.
I’d like Rihanna to be Vocaloided, because she’s such a musical talent.
With software like Vocaloid and EZdrummer you have more than sounds at your disposal, you can use the musicianship of the people that make it. The classical experience of Prima, or the clarity of mind of David Haynes on EZdrummer’s Jazz expansion pack. The more musicians realize this, the more music software producers will build their software around individual musicians, rather than around sounds or styles.