Why Does a Drum Sample Cut Off—or Keep Playing? A Trigger vs Gate Pad Test

Do not collapse four endpoints into “the sample stopped”

The MIDI Association documents Note On and Note Off as separate performance events. In a recording or sequencer, the interval between Note On and its matching release event commonly becomes the represented note duration, but MIDI does not force every receiving sampler to turn silent at that release. The instrument decides how its playback mode and envelope interpret the event. Effects can continue after the dry instrument has already ended.

“Trigger,” “One-shot,” “Gate,” “Sustain,” “AHD,” and “ADSR” are interface labels, not a universal naming standard. Ableton Drum Sampler, for example, continues according to Hold in Trigger mode but starts its Decay after release in Gate mode. Ableton Simpler’s One-Shot Trigger plays past release, while Gate begins its fade at release. Apple Ultrabeat disregards MIDI note length when its Sustain stage is off. Native Instruments Maschine distinguishes whole-sample One-shot, fixed-time AHD, and hold-sensitive ADSR. Compare documented behavior, not the label alone.

Freeze the source, note, and monitoring path

1. Choose a revealing source: use one clean sample with an audible body and tail. Set its playable source region so every N1–N4 release occurs before the source endpoint; otherwise the file can end first and hide whether release matters.
2. Save a control copy: duplicate the preset or instrument. Do not overwrite the only working version while testing modes.
3. Freeze the event: use one project tempo, MIDI note number, channel, pitch, relative start point, sample end point, and velocity. Turn looping off unless looping is the specific problem.
4. Start dry: bypass reverb, delay, feedback sends, and processing that changes the source decay. Keep the audio path, buffer, and output level fixed.
5. Leave room after release: record or render beyond the source’s expected tail. Otherwise a clip boundary can impersonate a sampler cutoff.

Do not trim the file, move its end marker, change pitch, and adjust the envelope in the same pass. Those changes can all alter audible duration for different reasons. This test earns one conclusion only when one layer changes at a time.

Probe the controller first: S for short tap, L for long hold

Record two physical pad actions at the same intended strength: S is the shortest clean tap you can repeat; L is a deliberate hold. Do not judge the audio yet. Open the recorded MIDI or a trusted monitor and write down Note On time, Note Off time, note number, channel, and velocity for both.

• S and L have different note ends: the input path represents release duration. Continue to the receiver test.
• S and L show the same note length: the capture path may normalize pad duration or the gesture may not produce a hold event. Do not blame the sample mode yet; use drawn notes for the receiver test.
• Note On appears but no matching note-end/release event appears: stop and fix the MIDI path. A sustained receiver can leave a note playing when its release event never arrives; monitors may display an equivalent MIDI release in different forms.
• Pitch or channel differs between takes: repair mapping first. Different receiving slots cannot isolate release behavior.

A physical drum pad may be designed primarily as a hit surface rather than a hold-sensitive key. The editor probe below deliberately removes that hardware question. Passing the editor test does not prove your controller transmits variable holds; passing S/L does not prove the sampler listens to them.

Draw N1–N4: four lengths, one start, one velocity

Create four separate test clips or four renders so tails never overlap. Each contains one note at the same project tempo, relative musical start, MIDI note number, channel, and velocity. Only the end moves: N1 = 1/32 note, N2 = 1/8 note, N3 = one quarter note, N4 = one half note. No universal BPM is required, but all four must use the same tempo, and the source region must extend beyond N4’s release. Render each long enough to capture the complete dry and later wet tail.

Listen, but also inspect waveforms or event data. Record a simple result code for each render: I = dry t2 is the same across N1–N4 and independent of their note ends, F = source reaches the full sample-region end, C = it cuts close to the note-end event, R = it begins a release fade there, U = unclear. A version can be both I and F; I can also end earlier at a fixed Hold/Decay value. “Sounds shorter” is not enough if a clip boundary or effect return was different.

Run the receiver in a fixed order: B0 → T → G → E → W → Q

1. B0 · current baseline: render N1–N4 without changing the loaded preset. Save the mode and envelope values, not just the audio.
2. T · release-insensitive candidate: select the documented state in which your product plays to a full or fixed endpoint without following note release—for example, the specifically documented Trigger, One-shot, or AHD behavior in the sources below. Render N1–N4 again.
3. G · release-sensitive candidate: switch only to the product’s documented state that makes note release relevant—for example, its specifically documented Gate, Sustain-enabled envelope, or ADSR behavior. Keep all four note events untouched and rerender.
4. E · envelope pair: while the release-sensitive state is active, make E0 with the shortest usable release/decay and E1 with a clearly longer one. Compare N1 and N4 first.
5. W · wet return: restore one effect or send at a time. Compare dry t2 with wet t3; do not call an effect tail a failed gate.
6. Q · quit/reload: save under a new name, close the instrument or project, reopen it, and repeat N1 plus N4. A setting that does not survive reload has not passed.

Not every sampler offers both T and G. If yours documents only one behavior, do not invent an equivalent control. The test can still identify whether note length affects audio, whether the envelope shapes the response, and whether the chosen state persists.

Use the first repeatable pattern to choose the next change

Do not remove note-end/release events to force One-shot behavior. A receiving instrument may ignore them for one mode and require them for another; other notes on the same path may sustain indefinitely without valid release events. Configure the receiver for the musical job and preserve well-formed MIDI.

Choose behavior by role, then regression-test the groove

Short kick, snare, or closed-hat events often make a useful control case for full playback, while a long texture or sustained instrument can expose variable Gate lengths clearly. That is a test-design choice, not a rule that every drum must be One-shot or every long sound must be Gate. After the isolated version passes, put the sound back into the exact groove that first exposed the problem and record three uninterrupted loops.

Check the original failure and two regressions: single hits still start correctly, repeated hits behave intentionally, and the saved project reopens with the same state. If changing mode fixes the cutoff but creates unwanted overlaps, do not compensate with several unrelated edits. Preserve both versions and run a separate retrigger/voice or choke test.

Is Note Off the same as muting audio?

No. It releases a MIDI note. The receiving instrument’s mode and envelope decide whether audio stops, fades, or continues to a fixed end.

Why does a sample still cut off in One-shot?

Check the sample region, fixed hold/decay, a second retrigger, monophonic voice behavior, choke groups, clip boundaries, and output effects. “One-shot” does not guarantee that every other layer is unlimited.

Why does changing MIDI note length do nothing?

The active mode may deliberately ignore Note Off, or the physical pad path may record fixed-length notes. N1–N4 separates those cases.

Is Gate always better for hi-hats?

No. Gate can provide variable release, but closed/open relationships may also depend on choke or exclusive groups. Test that pair separately rather than assuming one mode solves both jobs.