Hello,
I've done a similar work before as part of an application that identifies which music is playing. We can use FFT or auto-correlation, but I recommend the later as it gives more accurate results with voice samples.
If you hire me I would deliver a fully commented class which receives an arrays representing PCM audio and returns the pitch of the sample. If you have a different interface spec, I would use it instead.
It would take me two business days to finish the job and I require no upfront payment.
Best regards,
Diego