If you're measuring things on a multimeter all you're getting is the RMS voltage, and that's in a very narrow frequency range, if you're measuring above the frequency your meter is good for you're going to get a distorted reading. Even the best RMS multimeters are usually only good into the low KHZ range, mine is only good up to a couple hundred hertz. If your meter isn't a true RMS meter then the readings you get can be pretty much discarded as useless as the internal circuitry will affect it's reading more than the signal itself.
Depending on how the 555 circuit is built changing the duty cycle will affect the frequency a bit, not much you can do about it on a 555, it's an analog device, if you want a more precise drive frequency and duty cycle you can get micro controllers that have glitch free PWM outputs that will give you very high precision frequency and duty cycle.
If you use a capacitor across the motor leads in series with a diode to prevent it from discharging to ground during the off cycle it will provide hold up power and smooth out the abrupt changes in voltages from a PWM source. When a discharged capacitor is connected it will suck a LOT of power as it's a low impedance load to a square wave so it will slow the sharpness of the power rise to the motor as it's own impedance will draw less power, then when the PWM is turned off the capacitor will be isolated by the diode and discharge it's available power to the motor smooth it's shutdown voltage by drawing power from the capacitor. Making the curves less sharp will decrease the noise the motor can generator, as a hard square wave will generator many odd harmonics well above it's natural frequency into the audible range. This would probably most benefit from a low PWM frequency, in the hundreds of hertz range. But the capacitor and diode will introduce some loss which need to be taken into account.