Experimentation is key. It's really not my intention to tell you how to use your equipment here. Rather, I'd like to cover a few important topics and offer some generalized suggestions on approaches to try when all else fails.
This section is written with the presumption that the device in question has been correctly identified as something that outputs a signal.

Protect your valuables! I don't think I've mentioned this yet, or if I have it wasn't clear enough to make me comfortable. Old tube apparatus can output impressive levels of voltage. Since the context of conversation here is more or less that of repurposing equipment for use in an audio application, you cannot expect the output to be usable or friendly as is.

To illustrate; several of my devices have easily accessible front panel mounted sockets that output adjustable DC voltages in excess of 100 volts, with an adjustable AC voltage on top of that also capable of an excess of 100 volts. My DVM tends to ignore DC voltages while I'm measuring in AC mode, which is where the signals live. Individuals familiar with biasing will not find this an alien concept, but an electronically illiterate musician type who is just diving in may only test for the AC component, turn that down to a reasonable level and then wonder why their amp has blown or mixer channel has ceased to function after plugging it in.

In short, know and understand the output capabilities of your device before risking any equipment downstream. DI boxes or capacitors in line with the signal should block DC, though be sure that you remain within the voltage/current rating of the part. DI boxes or transformers can be tricky, as some don't tolerate DC current as well as others.

Taming high frequency. It should come as no surprise that the vast majority of apparatus that does not tout itself as useful at audio ranges is more or less useless for audio. After all, these things were built in the days of radio and television, where superiority in supersonics was another pinnacle of human endeavor. Signal generator is a fine example, since the term signal hints at something useable without actually quantifying a frequency. It's a bummer buying one of these beauties and rabidly waiting while someone drags their ass on shipping, only to discover several weeks later that it doesn't live up to your hopes and expectations of being the least bit usable.

So, you're going to find a wide selection of beautiful examples of early electronics that will not do a fucking thing for you when you plug it in, even ones that clearly have an input jack labeled "audio". Here's what's going on; they happen to be operating at a frequency beyond human hearing. Sometimes a frequency so high that even equipment which boasts flat frequency response beyond 100 khz doesn't pass it.
That audio input, if it has one, is there in order to modulate the carrier (high frequency) to simulate an analog radio broadcast at the frequency for which the device is set. A receiver, set at this frequency should output the audio that's been fed into the signal generator.

There is another method of wrenching a usable signal out of RF range oscillators, though it requires two sources of RF, preferably at least one which will allow for fine tuning; a modulator. I'm going to guess that you know of the ring modulator, which was itself birthed as an effect from the "de"modulation stages of AM radio receivers.

At this point, we're well on our way to Theremin territory, and all you need to do to complete the pitch function of the Theremin is to displace the variable tuning capacitor on one of the RF generators from a panel mounted component to a circuit that responds to capacitance created by the proximity of your body to an electrode.

Another tactic in detuning an RF generator to deliver audio range signals is to change (typically increase) the values of the components that comprise the tuned circuit that drives the oscillator. The actual process will vary from unit to unit, and I just mention it for completion.

Sync. Sync is sort of a brute force method to over-ride the panel setting on one device and make it fall in line with another. The way this is usually achieved is a pulse or sawtooth signal in the 20-30 volt range, fed into an astable stage of the "slave" oscillator. I find that typically the controls of the slave device are not disengaged and can occasionally provide interesting results at certain settings, though this varies wildly from device to device and set-up to set-up.

I should reiterate the voltage levels of sync do not lend themselves for direct use as audio, and feel compelled to add that I have come across sync levels that are over twice the 20-30 volt range stated above, and are ample enough to really let you know of their presence should you happen to physically graze an uninsulated contact.