You might be able to use a laser, with a video camera as the sensor. Even though the frame rate of the camera is much less than the sampling rate required for audio, it potentially has a high enough spatial resolution that there are multiple pixels per audio wavelength, giving enough bits of information per unit time. Some smart maths, effectively modelling the ribbon and matching its response to the observation, would be required. In this way, the vibration of the whole ribbon might be measured.