Just a couple of notes on generation of clean sin wave from my limited experience.
There are a few methods on generating high purity signal with equipment that isn't good enough. There are a few papers on this topic which you can find using google, but most solutions are too complicated IMO with much calibration, so I would not go there unless there's no other way.
Harmonic distortion is something you can typically solve using filters. Getting very good filters can be expensive though, anywhere from 10's of $ a piece to a few 1000's. They will also only give you a limited frequency coverage obviously, so you'll need a few if you want to measure a few frequencies. There are various filter types depending on the frequency of interest and the tolerance, 3dB BW you are looking for, and signal power as well as the impedance of the line.
For audio use, you can probably roll your own filter without too much effort or cost. You can verify its transfer function vs. frequency to make sure it meets your needs to get a sufficiently clean signal at the output.
One thing which is often overlooked but can be an issue is non-harmonic spurs.As part of building a test setup for something at the office, I've made a few measurements a couple of months ago testing the spurs of the SDG2082X we have there. The measurement is done at 15Mhz due to the old HP generator I've used for comparison in the measurements I'm attaching, but I've seen similar results at the higher frequency too. Harmonics weren't too important for this use case since we have a very sharp LC filter placed after the generator which cuts it all done, its the spurs that are within the pass-band of the filter that I wanted to measure here, therefore the limited span.
Measurement is 5Vpk-pk sine-wave at 15MHz into 50ohm termination of the Siglent SA. The figure is attached below, green line is the SDG2082X, in blue is the SDG1020 I have at home, and in pink is my old HP 33120A. In this case both Siglent generators show spurs of similar level (the SDG1020 is a tad lower, but closer to the center frequency), but the HP is much cleaner which is very disappointing considering how old (and slow) it is compared with these more modern instruments. I've assumed that the 2082X will be significantly cleaner in this regard.
BTW, I did measure distortion of the generators too, but didn't compare. They were all within spec, but I didn't keep the results unfortunately.
The spur level in the figure is quite low and is within the spec of the instrument obviously. But the emphasis here is that this is something you can't filter out as easily as harmonic distortion.
If relevant for you
, non-harmonic spurs can be more of a problem than harmonic distortion. These can be very close to the signal of interest, so filtering them out will be more difficult. You would typically need a spectrum-analyzer to find them, but for audio you can use a sound card as a spectrum analyzer, there are plenty of great programs out there to help you.