How to Measure Timing Differences When Retriggering a Still-Playing Sample
To measure retrigger timing, record a precisely scheduled pair of sample triggers, locate the same waveform landmark for each hit, and compare their measured spacing with the scheduled spacing. Then compare the second hit with an isolated playback at the same timeline position. This separates an overall timing offset from a change associated with retriggering while the sample is still playing.
This workflow is for producers testing a sampler in a DAW. It uses Ableton Live’s Simpler as a documented example and Audacity for numerical waveform measurements, but the comparison method applies to other samplers. The goal is a defensible measurement in audio samples and milliseconds, with enough controls to explain what it means.
What timing difference are you measuring?
Start by choosing the quantity you need. These three measurements answer different questions:
| Measurement | How to calculate it | What it answers |
|---|---|---|
| Recorded onset spacing | Second onset position minus first onset position | How far apart did the two audible starts occur? |
| Spacing error | Recorded onset spacing minus scheduled trigger spacing | Did the recorded hits preserve the programmed interval? |
| Additional second-hit offset | Second onset in the paired take minus its onset in a matched isolated take | Did the second hit move when an earlier voice was present? |
A fixed delay affecting both hits cancels when you subtract their onset positions. Consequently, correct spacing does not establish zero trigger-to-output latency. Likewise, a difference between two rendered files describes those renders; it does not establish the delay between pressing a controller and hearing the result.
Use the additional second-hit offset when your main question is whether retriggering an active sample changes its timing. Use spacing error when your practical concern is the rhythm of repeated hits.
How does the sampler handle a sample that is already playing?
Confirm the playback mode before recording. Restarting a voice, replacing an older voice, and adding another overlapping voice produce different waveforms.
In Simpler’s Classic mode, Retrig cuts off an already sustaining note when that same note plays again. With Retrig disabled, copies of the same note can overlap. The manual also conditions the audible effect of Retrig on a long release and more than one available voice. One-Shot mode is monophonic, and its Trigger setting lets playback continue after note release. These controls describe different aspects of playback. See Ableton’s Live Instrument Reference.
Record the mode, voice count, Retrig setting, note lengths, release, sample start position, pitch, and any fade settings. If you are testing overlapping playback, verify that the first voice actually remains audible when the second trigger arrives. A long source file alone does not prove that the instrument’s envelope keeps it playing.
How do you build a repeatable retrigger test?
Use programmed notes rather than manually striking a pad. Begin with one fixed note, identical velocities, and a sample with a recognizable opening followed by a sufficiently long tail. Choose a trigger interval shorter than its active playback duration.
Create three versions of the same arrangement, keeping the timeline origin and export range identical:
- First-only control: the first trigger plays; the second is absent.
- Second-only control: only the second trigger plays, at its original timeline position.
- Paired test: both triggers play, with the first still active when the second arrives.
Keep all sampler settings identical between versions. Preserve the initial silence in every recording or export. Automatically trimming each file to its first sound would remove the shared timing reference needed for the comparison.
For a baseline, remove unrelated processing and disable intentional timing variation such as groove or humanization. Use a fixed sample start and pitch. If you later test a musical configuration, restore one setting at a time. In Simpler, disabling Warp makes playback speed change with pitch, so changing notes would introduce another variable; the instrument manual explains that behavior.
Keep routing and delay-compensation settings fixed. Ableton documents that devices and track delays can introduce latency, and that Delay Compensation aligns tracks by offsetting them. It also states that graphics are not compensated, so a meter animation is unsuitable as the timing reference. Measure the recorded audio instead. See the Delay Compensation FAQ.
Choose either internal recording or export for the initial comparison and use it for all three versions. Testing real-time hardware playback is a separate measurement path; do not mix that result into a comparison of internal renders.
How do you locate the second onset inside an existing tail?
Import the files with their shared origin preserved, use one analysis sample rate, and examine the isolated controls first. Identify a repeatable landmark near the start: for example, the beginning of a distinctive short waveform pattern.
Write down the landmark rule before measuring the paired take. “The first sample above a chosen amplitude threshold” is usable for an isolated signal if you document the threshold. It becomes unreliable when the preceding tail already exceeds that threshold. The largest nearby peak is also a poor default because it may belong to the combined waveform rather than the new hit.
For overlapping voices, compare the paired take with the first-only control. Under the explicit assumption that the first voice remains unchanged and the signals combine linearly, subtracting the first-only waveform from the paired waveform reveals the second contribution. This is an analytical control derived from that assumption, not a documented guarantee about a sampler.
Do not use that subtraction as an isolated-second reconstruction when retriggering cuts off or fades the first voice. In that case, the difference also contains the removed or altered tail. Instead, compare a recognizable portion of the restarted waveform with the second-only control, after any replacement transition. If no unique landmark remains visible, report the onset as unresolved or give a bounded range.
If necessary, repeat with a diagnostic sample that has a distinctive opening and a long tail. Treat that as a new test: it establishes behavior for the diagnostic sample and settings, while the original musical sample may still require separate inspection.
How do you read exact sample positions in Audacity?
Set the selection readout to samples and zoom closely around each chosen landmark. Record its position numerically rather than estimating milliseconds from the horizontal ruler.
Audacity’s Selection Toolbar supports sample-based formats, editable selection boundaries, and a Start and Length display. Its manual warns that editing time values snaps them to the selected format, making the choice of units consequential. See Selection Toolbar.
Toolbar names vary by version. The newer Audacity manual describes a Bottom toolbar with editable selection start, end, and duration, including a samples format. Use the controls corresponding to your installation; see Bottom toolbar.
Measure the first and second landmarks, then subtract their sample positions. Alternatively, select between them and read the selection length. Check both boundaries after adjusting a selection so an accidental boundary move does not enter the calculation.
The time conversion is: difference in milliseconds = difference in audio samples × 1,000 ÷ analysis sample rate. Here, “audio samples” means discrete values on the recording timeline, rather than the sound file loaded into the sampler.
What does a worked retrigger measurement look like?
The following numbers are illustrative, not measurements of a particular product. Assume a 48,000 Hz recording and scheduled triggers 12,000 audio samples apart, equivalent to 250 milliseconds.
| Item | Position or difference |
|---|---|
| First onset in paired take | 48,096 samples |
| Second onset in paired take | 60,144 samples |
| Second onset in second-only control | 60,096 samples |
| Recorded onset spacing | 12,048 samples |
| Spacing error | +48 samples |
| Additional second-hit offset | +48 samples |
The recorded spacing is 12,048 ÷ 48,000 × 1,000 = 251 milliseconds. Compared with the scheduled 250 milliseconds, the spacing error is +1 millisecond. The second hit also appears 48 samples later than in the second-only control, supporting an additional second-hit offset of +1 millisecond under these test conditions.
That interpretation depends on using equivalent landmarks and preserving alignment. If overlap makes the paired landmark uncertain by roughly the size of the reported difference, the result needs a clearer diagnostic signal before it supports a timing conclusion.
How do you distinguish a consistent offset from variation?
Repeat the three-take comparison several times without changing settings. Log the paired onsets, isolated second onset, calculated differences, and any uncertainty in identifying the landmark.
A stable additional offset suggests a repeatable difference under the tested conditions. A changing offset calls for further isolation. Change one variable at a time: trigger interval, playback mode, available voices, processing chain, or capture method. Retain the controls for each configuration.
Also test an interval long enough for the first voice to finish. If the difference occurs only at shorter intervals, that helps associate it with active playback or voice handling. It does not, by itself, identify the sampler’s internal implementation.
Finish with a scoped result: “At 48 kHz, with these playback and routing settings, the second onset was this many samples later than the isolated control, across this many repetitions.” Keep the recorded files and landmark rule with that statement. This gives you a reproducible basis for adjusting the production session or investigating the sampler further.