Sure, a battery solves the noise problem and a 9V certainly offers a lot more headroom. I did prototype a solution using a battery. Since I didn't want to have the battery drain when not in use or require the user to unplug the input jack like some pedals do, I created a two transistor switch which turns on the buffer when the pi receives power. That worked great, dead quiet, but it seemed odd requiring the user to add (and eventually replace) a battery to a pedal that has external power.
The buffer circuit I used is from this handy tutorial about guitar buffers:
I tried most all of them, but settled on the second from the last. A simple Op amp with voltage-divider biasing. The TL071/TL072 won't work with 3.3v and somewhat noisy even with properly powered, so I just swapped it for the MCP6292-E/P. I also changed R1/R2 to 2Meg for an input impedance of 1Meg. Also, since the AudioInjector card has 10uF input caps, I didn't need that on the output of the buffer. The MCP6292 is a dual opamp, so you get 2 channels per 7 parts (4 resistors, 2 caps, 1 opamp). Not bad.
For the power regulation, I just use the AMS1117-3.3. Ground the adj terminal. A 220uF (could probably be as small as 10uF) and 0.1uF bypass ceramic across the 5V input. A 10uF oscon across the 3.3v output and another 0.1uF bypass ceramic near the opamp. Standard regulation. I believe the Audioinjector boards include an inductor on the output to form a simple LC low-pass filter. 13uH should roll-off at 14kHz. 6uH at 20kHz so somewhere in there should help, but it didn't seem to make a noticable difference so I left it out.
Regarding a commodity widget. Tons of guitar effects include an input buffer similar to those AMZ presents. You can also buy little boards that do that:
So I'm guessing maybe you're creating an audio interface for recording? I'd be interested in hearing what you come up with.
Good luck!
-Rand