We study a resonator coupled to a generic detector and calculate the noise spectra of the two sub-systems. We describe the coupled system by a closed, linear, set of Langevin equations and derive a general form for the finite frequency noise of both the resonator and the detector. The resonator spectrum is the well-known thermal form with an effective damping, frequency shift and diffusion term. In contrast, the detector noise shows a rather striking Fano-like resonance, i.e. there is a resonance at the renormalized frequency, and an anti-resonance at the bare resonator frequency. As examples of this effect, we calculate the spectrum of a normal state single electron transistor coupled capacitively to a resonator and of a cavity coupled parametrically to a resonator. PACS numbers: 85.85.+j, 85.35.Gv, 42.79.Gn When a mechanical resonator is coupled to a detector, even weakly, the detector can have a significant effect on the dynamics of the resonator, and it is this "backaction" that ultimately enforces the standard quantum limit of measurement [1]. For weak enough coupling, the detector acts like an additional thermal bath, providing an effective frequency shift, damping and temperature. This effective temperature can be lower than that of the resonator's environment, so a detector, such as an optical/microwave cavity or a mesoscopic conductor, can cool the resonator [2, 3], potentially to its ground state ([4] and refs. therein). Linear response theory gives a general way of calculating the noise spectrum of a resonator coupled to a detector, which is found to be very close to a thermal spectrum [5]. A relevant question is if a thermal model is enough to fully capture the dynamics of the system, or if there are any effects beyond a purely thermal back action. In particular, when calculating the spectrum of the detector, can we still treat the back action as purely thermal?In this Letter, we consider a detector linearly coupled to a resonator, and calculate the noise spectrum of both. As expected, the resonator spectrum is essentially thermal, with the frequency, damping and temperature modified by the back-action. We might therefore expect that the detector spectrum is close to the spectrum of a backaction-free detector coupled to a resonator with this modified thermal bath. However, we find that this is not the case, and show that the noise in the detector instead has a rather striking feature akin to a Fano resonance [6], i.e. a resonance at the renormalized resonator frequency plus an anti-resonance at the original frequency. Fano resonances arise from the interference between coherent and incoherent paths in mesoscopic conductors [7], but resonance/anti-resonance pairs are a general interference phenomenon, occurring in systems from LC circuits to electromagnetically induced transparency (EIT) [8,9].We first outline our general formalism before deriving an expression for the spectrum of a generic detector coupled to a resonator. Finally, we illustrate the analysis with two examples: a single electron tr...