If we have to find the speed of sound in any material measuring sound vibrations at different points in the sample, what kind of piezoelectric sensors which have high sensitivity would be useful?
For the measurement of sound propagation speed in different material is PVDF and Lithium Niobate piezoelectric materials. For the contact method generally a disc type transducers are used. The resonant frequency of interest (which results in resolution)is defined by it's thickness and diameter. Other methods are also available which are non-contact.
I would add that Electret microphones the size of a button are incredibly inexpensive and have a frequency response in some cases from 10 Hz to 30KHz. These are used to monitor earthquakes by the USGS in the U.S. and are very reliable.
Experimentalists state that to generate and detect acoustic waves in the piezoelectromagnetics (PEM), also known as the magneto-electro-elastic materials, is more preferable with such non-contact method as the electromagnetic acoustic transducer (EMATs). Also, such the two-phase materials possessing both the piezolelectric (PE),and piezomagnetic (PM) effects as well as the magnetoelectric effect can be used for smart materials. Anyway, PEMs can be used instead of PE and or PM. PEMs can give more possibilities for sensors, see my books for the non-dispersive waves' propagation in the PEMs and I have also just written one book on the thirty two new dispersive SH-waves' propagation in PEM plates. It is stated and well-known that SH-waves can possess a larger sensitivity.
You can use piezoelectric sistem with quartz resonator. Shifts of frequency and shift of q-factor of the system allow you to measure shear viscoelastic properties of material (mainly fluids).
I would concentrate on coupling considerations. How to get proper acoustic match on the contact of transducer vs. sample surface. What kind of materials are you measuring? The speed of sound is quite easy task if you need only longitudinal. 1 MHz transducer will probably be enough, and you get good penetration in usual solids. I only have experience of Olympus-IMS US (formerly Panametrics) transducers and Vallen-Systeme emission sensors. Both work fine.
What kind of samples are involved? Piezoelectric crystals are usually narrowband so if the sample is bound to exhibit dispersion or anisotropy, you may have to 1) use multi-frequency techniques (e.g. pulses or several piezoelectrics tuned to different center frequencies), or 2) excite body resonances to infer sound speeds of various modes and polarizations. Unless you'd be satisfied with an average or ballpark figure for sound speed, the techniques differ whether your sample is crystalline, amorphous, porous, granular, with magnetic domains etc.
Since you are coupling to steel, none of the transducers which are based on flexing a membrane are useful (ie gas and liquid sensors designs are not good). You would probably do best with a PZT ceramic element in a rigid matrix (epoxy) within a faraday shield. The PZT and steel should have similar acoustic impedances.