Why Your Ostinatos Sound Like a Machine Gun

Ostinatos are where sample libraries get caught. A sustained line hides a lot, because the ear is listening to the shape rather than the sample. A repeated quaver figure gives the ear nothing to listen to except the sample, twelve times a bar.

And once you have heard the machine gun in your own cue, you cannot stop hearing it.

Count Your Round Robins Before You Write

A round robin is a second, third or fourth recording of the same note, cycled automatically so consecutive repeats are not identical. Most orchestral libraries have some. Fewer have enough.

Do the arithmetic before you commit to a tempo. At 150bpm, semiquavers arrive ten times a second, so a library with four round robins returns to its first recording every 0.4 seconds. That is fast enough that the ear hears the cycle as a pattern rather than as variety, and the pattern is what reads as mechanical.

Six to eight round robins pushes the return out past a second, which is roughly where it stops being audible as repetition. Libraries that have gone to that trouble say so, and libraries that stay quiet about it usually have two.

The friction is that the number is often undocumented and occasionally wrong. The honest way to find out is to play the same note twenty times at a constant velocity, bounce it, and look at the waveform for the point where the shapes start repeating.

Spread the Velocity, But Know What It Controls

The standard advice is to vary velocity, and it is half right.

It works when the library maps velocity to separate recordings, because moving velocity around pulls in genuinely different samples and multiplies your effective round robin count. It does nothing when velocity is wired to volume, which is common on short articulations, because you get one recording at several levels.

Find out which you have. Play the same note at velocity 40 and velocity 110 and listen for a change in tone rather than only in loudness. Tone change means layers, which means velocity is doing real work.

Where it works, spread across at least 20 velocity points and do not centre the spread. Real repeated playing drifts, it does not oscillate around a mean, so let the line wander up over four bars and back down rather than jittering.

Playing connects you to YouTube, which may set cookies.

Vary Length, Which Almost Nobody Does

Here is the one that costs nothing and gets skipped.

A player repeating a short note under pressure does not produce identical note lengths. The bow shortens as the passage tightens, and lengthens slightly when there is room, and that variation is a large part of what the ear reads as a human being working.

In your DAW, take the ostinato and vary note lengths by 10 to 20 percent across the phrase. Not randomly, in a shape: shorter through the crescendo, longer as it settles. On a sampled short this changes which part of the release you hear, which changes the tail of every note.

The annoyance is that it interacts badly with some libraries. A few short patches ignore note-off entirely and play the whole sample regardless of length, in which case this does nothing and you have wasted ten minutes finding that out. Test with two notes before you edit forty.

Find the Round Robin Reset

Every sampler has a rule for when the round robin cycle returns to position one, and the default is frequently wrong for ostinatos.

Kontakt libraries commonly expose a reset on transport stop, which is what you want: the cycle continues through the phrase and starts fresh when you do. Some libraries reset on every new note, or on a change of velocity layer, and either of those hands you the same sample far more often than the round robin count suggests.

Look for it in the library's own interface rather than in your DAW. It is usually a small menu labelled reset, cycle or RR, and it is usually not on the front panel.

The friction is real and specific: if the reset happens on transport stop and you are working in short loops, every loop pass starts on sample one, so the first few notes of your ostinato are always identical while everything after them is fine. You will hear the problem at the loop point and nowhere else, and chase it in the wrong place for an hour.

Alternate Two Libraries Across the Bar

When a passage genuinely will not sit, stop trying to fix one sample set and use two.

Put the ostinato on two tracks with different libraries, and alternate: bars one and three from the first, bars two and four from the second. Two different rooms, two different sections, two different sets of recordings, and the ear stops hearing a loop because there is no longer one loop to hear.

Keep the two at slightly different levels rather than matched, with the second maybe 1.5dB under, so the alternation reads as texture rather than as a switch.

The catch is reverb. Two libraries recorded in different halls will announce themselves at the swap unless you commit both to a shared reverb and pull their built-in room back, which is work you have to actually do rather than assume.

When to Give Up and Buy the Phrase

There is a point where the honest answer is that you are asking sampled shorts to do something they were not recorded to do.

Phrase-based ostinato libraries record the whole repeating figure as a continuous performance, so nothing retriggers and there is no machine gun to eliminate. For fast driving patterns at fixed subdivisions they beat anything you can program, and they beat it in ten seconds rather than an afternoon.

The trade is that you get the patterns the players recorded, at the speeds and in the keys they recorded them. That is a real constraint on the writing, and it is why these libraries live alongside a normal string section rather than replacing one.

Know which problem you have. A composed line with repeated notes needs the five fixes above. A driving rhythmic bed that is going under a trailer cue probably wants the phrase library, and reaching for it is not cheating.