We present IFS-RedEx, a spectrum and redshift extraction pipeline for integral-field spectrographs. A key feature of the tool is a wavelet-based spectrum cleaner. It identifies reliable spectral features, reconstructs their shapes, and suppresses the spectrum noise. This gives the technique an advantage over conventional methods like Gaussian filtering, which only smears out the signal. As a result, the wavelet-based cleaning allows the quick identification of true spectral features. We test the cleaning technique with degraded MUSE spectra and find that it can detect spectrum peaks down to S/N = 8 while reporting no fake detections. We apply IFS-RedEx to MUSE data of the strong lensing cluster MACSJ1931.8-2635 and extract 54 spectroscopic redshifts. We identify 29 cluster members and 22 background galaxies with z >= 0.4. IFS-RedEx is open source and publicly available.