Audio Voice detection accuracy. Spring 2019

Advice needed please - 1st off I am not a programmer, I am producing this project, so please be gentle! Unity voice detection issues.
Hi. We are developing an Android mobile app that detects peoples voices and singing notes. We are dogged with Latency, high CPU and variable detection success. Are there any specialists here that can advise? Kindest Regards Jason

Some further notes from my Unity designer:

Picking notes from a human voice in real-time is causing us issues with accuracy of note detection. The CPU usage is high causing the phone to heat up. The app starts to stutter and affects the mic making note detection variable.

We want our mic to be accurate in note detection using as minimal CPU as possible. We want very little (unnoticeable latency). We want parameters of the note detection to be set by the app (us).

Technical: the app is built in late 2018 version. Unity 2018 .1

We are only using only unity SDK’s (package) for voice detection.

At the moment it seems variable and sometimes random especially phone cpu is high.