How to create a tensor model for a 1 D time series raw data

I have an EMG data having two channels and almost 20000 samples. I want to perform denoising on the data. I have done it using wavelet transform. I want to apply some denosing algorithm, to a tensor, hoping to get a better result. But none of the resource are conveying much information of how a tensor can be formed based on some raw data like the EMG data.