One thing to watch out for when using sound cards for analysis work, especially cheap sound cards, is that you need to know what is caused by the sound card and what by your DUT (much less of a problem with proper gear). For example, the sound card in my laptop has usable distortion (about -90 dBc), but a perverted amount of jitter, which results in strong sidebands (first two sidebands at -50 dBc, plus minus a bit, depending on frequency). The issue is now: without secondary known-good tools, you can't really do such characterization. But a basic low-distortion sine generator can help around that, because you know that, for example, an SVO or Wien oscillator simply never shows jitter sidebands (since it's not a sampled system, obviously). With it you can characterize your ADC which you can then use to characterize the DAC. The ADC is usually better than the DAC, curiously. (You'd expect the focus would be on the DAC, but maybe it's just that a good DAC is too hard to implement in a fully integrated single chip HD audio codec.)
(Also, this characterization is not perfect, obviously, because in most integrated codecs the ADC and DAC are driven by the same clock, which can cancel some (but not all) forms of jitter.
"not any good for audio work" -> well of course it is. At least I work mostly in the time domain when working on amplifiers, "switching" to frequency domain mostly for characterization purposes. Also, a sound card won't ever see HF oscillations in any amplifier. So a scope is always worth it's weight in gold.