A guitarist can hear a delay while playing and still find the recorded part sitting exactly on the beat. Two timing measurements matter here. Round-trip latency is the elapsed time for audio to travel out through the interface and back into the computer. Placement error is the gap left on the recording timeline after the DAW has used the driver’s timing information. Measure both before applying an offset to every future take. Oblique Audio’s RTL Utility guide · Ableton’s recording-compensation FAQ
Build a line-level path that ends at the recorder
Save your work and note your normal interface routing. Mute or disconnect the monitors and turn headphone levels down. Pick one physical line output and one physical line input on the same interface. For example, connect Line Out 3 to Line In 2 using a cable appropriate to the two jacks. Set a switchable input to Line, disable phantom power on that channel or bank, and keep microphones disconnected from the test path. If the interface has no supported analogue line return, use a compatible interface for this exercise. A speaker amplifier output is unsuitable. Focusrite’s signal-level guide · Focusrite’s hardware and phantom-power guide
In the interface mixer, remove the line input from the mix that feeds the tested output. Disable direct monitoring and DAW input monitoring for the return. The only signal feeding that output should be the test signal generated by the computer. This breaks the potential feedback circuit while preserving the input recording. Start with the output attenuated and the line-input gain low. Oblique Audio’s physical patch procedure · Focusrite’s direct-monitor routing explanation
Measure the physical round trip
- Close the DAW if it holds the audio device. Open RTL Utility, choose the interface and its driver, and enable just the patched input and output. Select one sample rate and buffer size, for example 48 kHz and 128 samples. Confirm both values in the device panel.
- Use Measure RTL. Check that the utility recognises the return and reports a valid result. Adjust analogue levels cautiously if necessary. Write down both the measured round trip and the driver-reported round trip. Keep the units visible.
- Run three measurements at identical settings. Record each result. Then change only the buffer size, for example to 256 samples, and repeat three times. Keep the per-setting results together. A buffer-size calculation describes only one part of a path that also includes conversion and driver buffering. RTL Utility measurement instructions · Ableton’s breakdown of audio latency
Here is an illustrative result. At 48 kHz there are 48 samples per millisecond. A measured 432 samples therefore equals 9 ms. If the driver reports 384 samples, it reports 8 ms. The difference is 48 samples, or 1 ms. Record all three numbers. Entering the entire 9 ms as a manual correction in a DAW already compensating for 8 ms would count most of the delay twice.
Verify placement inside the DAW
Close the utility and reopen the DAW at the tested rate and buffer. Record your existing manual offset before temporarily resetting it to zero in a test session. Use the DAW’s own calibration procedure. In Live, open the built-in Driver Error Compensation lesson and use its prepared set. Keep input monitoring off on the return so Live applies its normal recording compensation. Route the source pulse through the same physical sockets, record the return and compare equivalent points on the source and recorded transients at sample resolution. Follow the lesson’s sign convention for any repeatable residual. Capture a fresh pass after changing it. Ableton’s out-of-sync recording procedure
If the return is absent, stop and check the channel numbers and cable before raising gain. If the measured delay jumps between passes, save the results and repeat using the single interface with plug-ins, virtual loopback channels and extra routing removed. A varying offset supplies no reliable single correction. If the DAW places the pulse correctly at zero manual offset, keep that setting. Oblique Audio’s RTL Utility guide
Make the listening check
Remove the patch cable before restoring normal monitoring. Play a short guitar figure against a click, once with direct monitoring and once through the DAW, using only one cue path at a time. Rate the feel during each take. Then listen to the recordings against the click without looking at the waveforms. You now have separate answers for how the cue feels and where the recorder places a known signal. Save the settings, three repeat measurements and fresh verification capture with the interface model and driver version. Recheck after a configuration change. Ableton’s recording-compensation FAQ
Related reading
- Build a bedroom studio around one clean take
- Diagnose crackles in a USB recording session
- Handbook contents and glossary