Mass wasting, such as rockfalls, landslides and debris flows in steep mountain terrain, has a high destructive potential, and plays a key role in both erosion and landscape evolution. As an alternative to many conventional approaches, seismology allows monitoring of such mass movements at safe distances, provides estimates of event location and timing, and can give insights into dynamics and rheology granular flows. Here, we analyze seismic data recorded during the 2017 and 2018 debris-flow seasons at Illgraben, a steep canyon located in Switzerland. Yearly precipitation is controlled by summer rainstorms with high rainfall intensity during which mass wasting including rock-slope failure and debris flows occur regularly. The frequent debris-flow occurrence (on average three events per year) makes the Illgraben an ideal site for cross-validating a seismically-derived event catalog of mass movements with "ground-truth data", such as digital terrain models, flow depths estimates and other in-torrent measurements. We present seismic frequency characteristics of the Illgraben debris-flow series and investigate how the seismic signature depends on actual debris-flow characteristics, such as grain sizes, and on propagation effects of the generated seismic waves. Whereas these two effects are usually difficult to separate, the source component contains valuable information on the flow’s material composition. Stations that are close to the torrent, we find that dominant frequencies in the recorded signal reflect the distance to the dominant source. For one particular station, this is shown on recordings of several events, where a dominant frequency of about 5.5 Hz indicates the passing of the flow at a 48m check dam. Power spectral densities at that instance give an estimate of the particle content of the debris flow. We also find that a jump in dominant frequency does not necessarily reflect the location of the flow front. Seismic studies of debris-flow dynamics and material composition should therefore not be limited to entire debris-flow seismograms, but instead focus on individual time windows and consider different sensors separately. The presented analysis underlines the use of seismic data in torrent and landscape studies.