OPA134 is indeed a very good (audio) opamp. For ultimate performance check out the OPA627 (very expensive) and this link:
High Speed Amplifiers (Greater than equal to 50MHz) Product Home from Texas Instruments
Those are used for DSL applications and are orders of magnitude faster and less noisey than the typical audio opamp. I'm pretty sure you can get samples from TI for free. Also, some of these don't come in a DIP package, they're surface mount only, in this case the only way is to get an adapter socket, but soldering SMT components requires special equipment and a very steady hand.
Some time ago I built a discrete class A complementary Darlington preamp and used the OPA134 as the buffer. Noise was practically unmeasurable, but the problem I had was a nasty oscillation at very high frequencies I think it was ~ 170KHz. The only way I could get rid of it (in addition to bypassing with ceramic caps), was to solder it directly on the pcb, instead of using a socket. Another thing to keep in mind is that some opamps are not stable at unity gain, so make sure that's not the case in your circuit. Btw, I just finished cleaning and testing 16 modules on the 200B.
Noise with 16 channels assigned and with gain at 40dB was -96dBFS (A weighed). Doing a star ground mod can lower it ~12 more dB's!!
Just because a circuit doesn't measure as well spec-wise doesn't mean it won't sound good. Many people swear by the 5532, claiming it has a less harsh distortion than the "better" opamps. Same thing with the TL072 and others.