Abstract There have been two major difficulties that hinder the use of a loudspeaker–pipe system as a short-duration acoustic point source generator. They are the non-flat response of the loudspeaker and the reflection at the pipe exit. Here, a pre-distorted digitally synthesized input signal is used not only to yield a desired impulse waveform but also cancel its first reflection and hence all subsequent reflections from the pipe exit. Our experimental results demonstrate the feasibility of generating repeatable sharp impulses of various desired waveforms and durations as short as 0.5 ms.