Onset Peak And Duration

Understanding Onset, Peak, and Duration in Sample Design

When you load a sample into a sampler or design a custom instrument, the three things that actually matter for how it sounds in context are onset, peak, and duration. Most people gloss over these because presets exist and most library sounds are fine. But when you start layering, pitching, or arranging rapid passages, the difference between a professional result and a mess usually comes down to how those three parameters behave. Let me explain the terms quickly, then show you how to work with them practically. Onset is the exact millisecond where the sound becomes audible after a trigger. Peak is the highest amplitude point within that initial transient. Duration is how long the body of the sound lasts before it naturally decays or needs to be cut. Together they determine whether a sample feels tight, loose, artificial, or natural in a mix.

Onset Peak And Duration in Practice

The method here is straightforward but requires attention to detail. You start by importing your source audio into your sampler of choice. Most samplers have a way to analyze and display the waveform so you can see the transient shape. Zoom in to the sample level. Look at the very beginning of the waveform. The onset is wherever the waveform first deviates from silence. This is not always the highest point. Sometimes there is a fraction of a millisecond of noise before the actual tone begins — a breath, a pick scrape, a mechanical click. That is your onset region. You need to decide whether to keep it or trim it. If you keep it, the sample will feel more organic. If you trim it, the sample will trigger more precisely, especially when playing fast passages or using quantization. Peak is the loudest point in that initial transient. In most drum samples and plucked instruments, the peak happens within the first 10 to 50 milliseconds. After the peak, the amplitude drops off into the sustain or decay portion. Your job is to make sure the peak is where you expect it and that it is not clipping. If a sample peaks above 0 dBFS, you will get digital distortion unless you apply gain reduction. Simply lower the gain by the amount that is clipped. A peak at minus 0.3 dB is fine. A peak at plus 1.2 dB needs correction.

Duration is the final decision. How long do you want this sound to last? In a sampler, you set this with the release parameter or by trimming the tail. The problem most beginners encounter is leaving too much tail on percussive samples. A piano note might ring for three seconds naturally, but if you are stacking six piano layers, those tails add up into mud. Trim the duration to what is musically necessary. For kick drums, you often want 80 to 200 milliseconds depending on the genre. For snare hits, 200 to 400 milliseconds. These are not rules, just starting points that work in most situations. Here is a specific problem I ran into last year. I was working on a sample pack for a Kontakt instrument and noticed that certain piano samples were phasing against each other when layered at different pitches. The issue was not the pitch itself but the onset timing. Because the samples were recorded with slight natural delays between the hammer strike and the string resonance, the onsets were not aligned across the keyboard. When you play chords, those micro-delays create comb filtering that makes the sound thin and hollow. The workaround was tedious but effective. I exported each sample, zoomed into the waveform, and manually adjusted the onset position so that the initial transient of every key started at exactly the same millisecond across the entire sample. I used a utility plugin that lets you shift audio without changing pitch. It took about two hours for a 61-key set, but once done, the phase alignment was tight and the layered sound became noticeably fuller and more coherent. I now do this as a standard step for any pitched instrument sample pack I produce.

There are a few counter-intuitive things about onset, peak, and duration that experienced engineers know but beginners often miss. The first is that a louder peak does not always mean a better sound. In fact, aggressively maximizing the peak of a transient can make a sample sound harsh and fatiguing over time. The human ear adapts to sustained loudness, and a sample that peaks too hard will cause the rest of your mix to feel quiet by comparison, pushing you to turn everything else down until the balance feels wrong. The second insight is that duration and perceived loudness are unrelated. A short sample with a strong attack can feel louder than a long sample with a gentle onset, even if the peaks are identical. This is why kick drums in electronic music are often under 150 milliseconds yet dominate a mix, while orchestral strings at similar peak levels sit back in the arrangement. Your brain processes attack energy differently from sustained energy. Another pitfall is assuming that all samplers handle onset the same way. Some samplers automatically align onsets to the grid when you enable groove templates. Others ignore onset alignment entirely and play the sample exactly where it starts. If you are building a kit or a multi-layered instrument, you need to check your sampler's behavior and verify the alignment by eye, not by ear alone. A visual check saves hours of troubleshooting later.

Here is how I approach this workflow step by step. Import the sample. Set the playback position to zero and look at the waveform beginning. Mark the first audible deviation as your onset reference. Measure the peak amplitude and adjust gain if needed. Determine the musical duration and trim accordingly. Export and test in context — play it alongside other instruments at the intended volume level. If it sounds good alone but disappears or conflicts in the full mix, go back and adjust the onset trim or duration. This process usually takes five to ten minutes per sample once you are familiar with it, and it prevents most balance problems before they happen. One important limitation to be aware of is that onset alignment works well for monophonic or single-transient sources but breaks down with sounds that have multiple distinct attacks, like a guitar strum or a brass fanfare. In those cases, trying to force every sample to start at the same millisecond creates unnatural artifacts because the internal transient structure of the sound is being distorted. For polyphonic or multi-attack sources, it is better to leave the natural onset timing intact and instead manage the overall level and EQ to fit the mix. If you need a free tool for this work, Audacity handles basic onset trimming and peak adjustment without any cost. For more precise work, especially with batch processing, I recommend using the free version of Decentra or sampling with a DAW that has built-in waveform editing. The specific tool matters less than the discipline of checking each parameter intentionally rather than relying on how a sample sounds in isolation.

I also want to mention that onset, peak, and duration are not the only factors that determine how a sample performs. Time-stretching algorithms, bit depth, sample rate, and the original recording quality all interact with these parameters in ways that can undermine your careful work. A 16 kHz sample that you have perfectly aligned and trimmed will still sound worse than a properly recorded 44.1 kHz source. Start with good recordings whenever possible, then apply onset, peak, and duration adjustments as a refinement step rather than a correction step.

Quick Reference Values

Kick drums typically need onset trimmed to remove sub-bass rumble below 30 Hz, peak set between minus 3 and minus 1 dBFS, and duration kept under 200 milliseconds for most modern genres. Snare hits benefit from onset preservation for natural feel, peak around minus 2 dBFS, and duration between 200 and 400 milliseconds depending on the room sound. Hats and percussion should have onset as tight as possible, peak no higher than minus 3 dBFS, and duration determined by the musical context rather than a fixed value. Pianos and other pitched instruments require onset alignment across the keyboard for phase coherence, peak management to avoid inter-sample clipping, and duration trimming that respects the natural decay while preventing buildup in dense arrangements.