ASIO4ALL Latency Compensation for Recording
A take can land consistently late or early on the timeline even when playback is stable. ASIO4ALL's latency compensation settings let you account for delay the driver cannot measure internally. Run a controlled test before adjusting the values.

In this guide
What ASIO4ALL latency compensation fixes
ASIO4ALL latency compensation corrects a constant timing offset, where recorded takes land consistently early or late on the timeline compared with what you played along to. You measure the real round-trip delay with a loopback test, compare it with the delay the driver reports, and adjust the compensation in small steps until a recorded click lines up with its source.
It is a calibration tool, not a speed-up. It does not reduce the delay you hear in your headphones, and it cannot fix tracks that slowly drift apart. Knowing which problem you have saves a lot of trial and error.
For a podcaster recording a single voice, a small fixed offset may never matter. For a musician layering guitar or vocal overdubs against a drum part, even a few milliseconds can make a performance feel loose. That is where a careful calibration pays off.
Signs you need it
- Overdubs always sit slightly behind the beat by roughly the same amount.
- A recorded click track never lines up with the original, even with plug-ins removed.
- The offset changes when you change interface or buffer, but stays fixed otherwise.
How ASIO4ALL reports latency to your DAW
When a DAW records, it needs to know how long the signal took to travel out to your speakers or headphones and back in through the input. It uses the latency the ASIO driver reports to shift the new recording into place.
ASIO4ALL v2 estimates the delay introduced by the host and hardware path, but some of that delay sits inside the WDM driver and the hardware where the driver cannot measure it. The compensation controls adjust the timing reported to the host so the DAW can place recorded audio correctly. The publisher documents the method on its latency compensation and loopback measurement page.
Recording offset vs monitoring delay
Live monitoring delay is a separate issue. It depends on the buffer and any processing between the input and your headphones. Lowering the buffer reduces what you hear; latency compensation only changes where the finished take is placed.
| Symptom | Likely cause | What to adjust |
|---|---|---|
| Takes always land the same amount late | Unreported path delay | Latency compensation or the DAW's recording offset |
| Delay in headphones while playing | Buffer size and processing | Smaller buffer, fewer plug-ins on the monitored track |
| Tracks slowly move apart over a long take | Clock drift between devices | Shared clock or a single device |
| Timing changes with a heavy effect chain | Plug-in processing delay | Remove plug-ins from the measurement |
Identify a consistent timing offset
Use a fresh project with the sample rate, buffer and device you normally record with. Create a short reference transient such as a click. Human timing varies from take to take, so use a physical loopback to check the equipment path separately from the performer.
A heavy effect chain can add delay of its own. Keep plug-ins out of the first measurement so the result describes only the driver and hardware path.
Prepare the test project
- Create one audio track holding the reference click and one empty track armed for recording.
- Set the output of the click track to the line output you will loop back.
- Set the input of the empty track to the line input receiving the cable.
- Leave a second or two of silence before the click so the transient is easy to find.
- Save the project as a template so you can repeat the exact test later.
A sharp, short transient is easier to read than a soft or sustained sound. A single-sample spike or a short metronome click both work well, because the start of the waveform is unambiguous on both tracks.
Convert samples to milliseconds
DAWs often show offsets in samples. Divide by the sample rate and multiply by 1000 to get milliseconds. For example, an offset of 96 samples at 48 kHz is 2 ms, and 441 samples at 44.1 kHz is 10 ms. Being able to switch between the two makes it easier to compare readings across tools.
Tip: Zoom in on the waveform until you can see individual samples before reading the offset. At normal zoom a few milliseconds is easy to misjudge.
Measure round-trip delay with a loopback
The publisher's method connects a compatible line output to a line input and uses a loopback utility to compare the measured round-trip delay with the driver's reported delay. This removes the performer from the equation and gives a repeatable number.
Step by step
- Connect a cable from a line output on your device to a line input on the same device.
- Turn off input monitoring for the returning signal so it does not feed back.
- Start at a low output level and raise it gently until you see a clean signal.
- Run the loopback measurement and note the measured and reported values.
- Repeat the measurement more than once with the same settings to confirm the result is stable.
- Adjust the compensation in small increments based on the difference between reported and measured latency.
- Measure again and repeat until the difference is as small as you can get it.
Watch out: A loopback with monitoring left on can create a feedback loop that gets loud very quickly. Keep levels low and your speakers or headphones turned down until you have confirmed monitoring is off.
Read the round-trip number correctly
A single round-trip value describes both directions combined, output plus input. Do not apply the whole figure to one side unless the method you are following says so. Small, measured changes are easier to verify than one large correction.
Confirm placement with a recorded click
Once the numbers look right, confirm the result where it matters. Record a click back into the DAW through the loopback and check where the recorded transient lands relative to the source.
What a good result looks like
- The recorded transient starts at, or within a few samples of, the source transient.
- Repeated recordings land in the same place.
- The result holds after closing and reopening the project.
Change one control at a time
Some DAWs provide an audio recording offset of their own. If you adjust both the DAW offset and ASIO4ALL latency compensation, the two corrections stack and the result becomes hard to reason about. Change one compensation control at a time, and check the DAW vendor's documentation to see whether it already applies an offset.
Typical audio settings locations are Options › Audio settings in FL Studio, Options › Settings › Audio in Ableton Live, Studio › Studio Setup › Audio System in Cubase and Options › Preferences › Audio › Device in REAPER.
How buffer size and sample rate affect the result
The buffer contributes directly to the delay the driver reports. At 48 kHz, 128 samples is about 2.7 ms, 256 samples about 5.3 ms and 512 samples about 10.7 ms; at 44.1 kHz, 256 samples is about 5.8 ms. Those figures describe one block, and the full path adds hardware, driver and application delay on top.
Because the path changes with these settings, a compensation value measured at one buffer or rate may not be right at another. The buffer size guide and the sample rate guide explain how to choose stable values first. The publisher's buffer size page has more background.
Note: Settle on the buffer and sample rate you will actually record with before calibrating. Calibrating at one setting and recording at another defeats the purpose.
Retest when the recording path changes
Keep a note of the interface, input and output, sample rate, buffer size and final values. When any of those settings change, record another reference click before relying on the previous measurement.
Multi-device setups
The publisher says ASIO4ALL uses the largest respective compensation value across devices in a multi-device setup. Different devices can therefore leave different residual offsets, so check each input separately with its own reference click.
If two recordings slowly move apart, check the devices for clock drift. A fixed compensation value moves a recording by a fixed amount and leaves that drift in place. The device aggregation and clock sync guide covers drift, and the publisher's device aggregation page explains the shared-clock requirement.
What to record in your notes
- Device and the specific input and output used.
- Sample rate and buffer size.
- Measured round-trip, reported latency and final compensation values.
- Any DAW recording offset in use.
Frequently asked questions
No. It changes where the DAW places recorded audio. Monitoring delay depends on the buffer size and any processing on the monitored signal.
A physical loopback from a line output to a line input is the publisher's method. It measures the equipment path without the variation of a human performance.
Use one correction at a time. If both are active they add together, so change one control, measure, and only then consider the other.
Compensation corrects a fixed offset. Gradual drift usually comes from two devices with separate clocks, which needs a shared clock or a single device.


