BJT complements don't really match, they're just close enough. Different brands would probably be fine. If multiple (same type per side) are being used in parallel, depending on circuit design, it might be worthwhile matching them (equal Vbe at given Ic, Tj). But that's usually a pretty shite design that needs matched BJTs.
Conversely, replacements of any type that fits (package, pinout) and has equal or greater ratings and parameters (including fT, hFE, and SOA -- check the SOA plot, make sure DC/100ms curve includes the same voltage range), will most likely suffice. The main gotcha is, modern epitaxials and especially ring/perforated emitter types, can have much higher fT than their predecessors, most especially anything that was designed for ye olde hometaxial 2N3055, 2N3773, etc., a process that was obsoleted in the... late 70s, early 80s, somewhere around there.
Note this hazard includes 2N3055 etc. themselves, modern ones greatly outperform the originals; but they still have the same old dumb datasheets because they have to, it's a JEDEC registered part, "meets or exceeds" shall be taken seriously.
Usually, a ferrite bead on the base, or adding a series resistor or increasing the existing value, suffices to tame the higher fT parts. You might also consider reducing wiring length, if they are much distance from the PCB.
Substitution requires a little bit of design knowledge so a one-to-one suggestion isn't feasible, but often little more than a judicious choice, and maybe a few resistor values, is all that's needed.
There's also NTE/ECG, the equivalents are usually alright. You are getting a common-denominator part though (many types cross to one), and again if the fT is higher, that's the main thing.
Tim