Why Smart Delay?
We get a lot of question about why do our instruments work this way?
A major hurdle we really wanted to solve with our New Standard series of libraries was simple: be able to play normal melodic phrases in a pop/jazz style and have it sound realistic. It’s a simple enough idea, but very difficult to achieve. As we’ve all heard in the plethora of jazz/pop horn libraries over the decades now, it’s very difficult to get right. Why is this? The reason it is so difficult can be boiled down to: the countless tiny variations and changes that happen as the player moves from one note to the next. Basically, once the first note is played, as the player approaches the next note, all kinds of things happen to pitch, timbre, volume and idiosyncratic sounds that are different for every instrument (and every individual player!). These changes are different depending on things like the tessitura that the instrument is currently in, the dynamic at which it’s played, where in the phrase the note is, the length of the note and the interval they’re coming from as well as the interval they’re about to play.
The truth is: there is literally no way for the duration of any particular note's sample to be idiomatically correct unless you know what happens after that note finishes. Does it lead to a rest (silence)? Is it the middle of a long string of fast notes? Does it leap up or down? As you can see from the image below, taken from a random stream of connected legato notes. A flat generic sustain will never suffice for
Example of full-length sample with transitions on both the beginning and end of notes. Notice all the interesting, almost random, dynamic changes throughout the body of the note.
Isn't Modeling The Answer Though?
We'll be the first to admit that the burgeoning world of instrument modeling is very exciting and a very cool area of development. And we're looking forward to see where it goes. But why don't we go down that road?
The other truth is: if a musician is reading music, their eyes are always at least a note or two ahead of where they're currently playing. Because what they're about to play affects the way they're playing the notes right now. And even if they're not reading music, they are thinking or subcontiously aware of what's coming next. They don't conceive of the note at the exact instant they play it.
So, while modeling is very interesting and yields some cool sounds and a lot of fun for musicians who enjoy playing them, it will never be able to contain all those idiosynchratic things that actual players really do when they play in the heat of battle.
What you end up with is a saxophone (for example) that sounds like a real instrument, just not like any saxophonist I've ever heard. All the details are wrong.
The only way to get all those intuitive details, is to record a real player, playing in real-time, in the heat of the moment and extract the samples from that performance and then re-assemble them later when the end user plays a new phrase.
How Do We Record For Smart Delay?
The main additions in The New Standard method of sampling are:
1. We recorded different lengths of notes. 9 different lengths. Players actually articulate and sustain notes very differently depending on how long they play them. Moreover, standard sustain samples, where the player performs an isolated long note, are only realistic and useful once the notes get to a certain length. If these standard long sustains are used for these shorter legato notes, the result is a flat, life-less and altogether fake sounding representation.
2. We recorded legato in and out of every note. So much of what gives each note of a phrase it’s unique sound, and why it’s so hard to replicate, is the preparation for the next note. Players do so many small changes as the next note is coming. Not only that, but it’s mostly involuntary and happens without them thinking (especially as the rhythm gets faster), so asking a player “play an 8th note” in a sampling won’t even close to get you the breadth of possibilities of what that same player might play in any given situation.
- To illustrate it more clearly, for every pitch there is a sample with a note from a whole step to a whole step, and from a whole step to a 3rd, and from a whole step to a 6th. Conversely, there’s sample from a 3rd to a whole step, and from a 6th to a whole, etc, etc...
3. We recorded different samples for where they are in the phrase. To use the 8th note example again: if you had a musician play three 8th notes in a row. Even if they’re identical length, pitch, dynamic, they will be performed totally differently. And neither of the 3 samples would suffice and sound realistic as a replacement for the other two instances.