Another limitation is that a DFT-based song detector appears more reliable in theory than in practice. DFT shows the frequency content of a signal, but it doesn't truly understand the musical identity of a song. If the audio clip being queried is too short, too noisy, or taken from a not very distinct part of the song, the extracted frequency peaks may not be sufficient for reliable matching. Therefore, we tested the method with parameter sensitivity, noise robustness, sample reduction, and a failure case with a very short audio clip.