The phoneme dataset has 370,000 phonemes and massive amount of noise. It is probably (99% certain) the largest single-dimension time series classification dataset.The data in the original dataset are of un-equal length. In this website, we just use a small subset of the original dataset, and we interpolate the data to make all time series equal length by doing dimension-reduction.
For the orginal dataset, please refer to: http://www.cs.unm.edu/~hamooni/papers/Dual_2014/index.html