One major source of nonlinear distortion in analog-to-digital converters (ADCs) is clipping. The problem introduces spurious noise across the bandwidth of the sampled data. Prior works recover the signal from the acquired samples by relying on oversampling or on the assumption of vacant frequency bands and on the use of sparse signal representations. In this work, we propose a different approach, which uses two streams of data to mitigate the clipping distortion. Simulation results show an SNR improvement of 9dB, while the conventional approaches may even degrade the SNR in some situations.