In speechband compression or expansion all the frequencies of the band are divided or multiplied in a certain ratio, butwithoutchanging the time dimension. This last requirement makes the problem very difficult because, if we could neglect the time aspect, we could record speech and simply play it back at slower or faster speeds.I have tried to solve this problem by three different methods: (1) by transformation of the speech frequencies into an optical spectrum and retransformation by light modulation disk and photo‐tube; (2) by use of a string filter; and (3) by use of an ultrasonic cell as a storage device.