How to achieve the best latency between Inputs (touches) and outputs (mainly Audio)

Hi !

I'm working on a Musical instrument app (Unity + iOS).
I'm trying to achieve the best latency between the user touch and the audio actually playing (and, graphics of course).

Latency is a cumulative of the following steps (correct me if something is wrong or incomplete) :
1. hardware touch detection
2. OS event detection
3. Unity process (Input.touches)
4. App process and logic
5. Audio output (OS)
6. Audio output (hardware)

I worked on the step 5 (and 4), rolling my own Audio Unit...

How to achieve the best latency between Inputs (touches) and outputs (mainly Audio)