We propose a new class of methods for sub-sample time-delay estimation.
Method based on auto- and cross-correlation without interpolation and fitting.
The proposed approach was robust when applied to real-world signals.
Algorithm presented moderated computational complexity.