Nigel has some good points here!!! You don't need a lot of capability for audio work.
As long as audio is the end game, bandwidth isn't really important as long as it is somewhat more than the audio spectrum - up to about 20 kHz. But there may be some high frequency oscillations that need to be seen to be dealt with. 10 MHz will show these oscillations in most cases.
But just as soon as you hit digital, bandwidth requirements increase; quickly. I have a DS1054Z unlocked for 100 MHz operation. The other day I was looking at a clock signal generated by an FPGA project. I wanted to be certain it was the proper frequency so I planned to use the scope's internal frequency counter. This worked fine and measured about 60 MHz, as I expected.
But the waveform looked like a distorted sine wave. Well, duh! The square wave would be composed of the odd harmonics of 60 MHz so the 3rd harmonic would be around 180 MHz, the 5th around 300 MHz and the 7th around 420 MHz and, in my view, a square wave doesn't look good without at least the 9th harmonic, possibly more. My poor little 100 MHz scope was totally inadequate. Even my 350 MHz Tek 485 would only display up to the 5th harmonic. I would need a 500 MHz scope to get a reasonable representation.
Had I bought the 200 MHz Siglent I have been lusting over, it might have been able to display the fundamental plus the 3rd harmonic - still inadequate if I were trying to look at signal integrity.
Fortunately, all I cared about was the frequency of the waveform, not the waveform display.
In my view, these low bandwidth analog scopes are useless for modern digital projects. That's why I got rid of my 10 MHz Heathkit scope and moved to the 350 MHz Tek.
An FPGA output at 60+ MHz is unusual for me. I am normally working at 10 MHz and below for things like SPI. As a result, my 100 MHz scope will display the 9th harmonic and that 200 MHz Siglent would be even better.
So, if the goal is to stay forever and always in the audio spectrum, almost any scope is good enough. But just as soon as you start playing in the digital sandbox, bandwidth requirements explode.
https://www.mathworks.com/help/matlab/math/square-wave-from-sine-waves.htmlThe real spec to worry about is the scope's risetime. But this is tied to the bandwidth on the nameplate by a constant of 0.35 so, if you have have the bandwidth, you will have the risetime and vice versa. Any discussion of 'equivalent time sampling' is omitted... It's just voodoo and requires a repetitive signal.
RT = 0.35 / BW where RT (risetime) is in nanoseconds and BW (bandwidth) is in gigahertz. For 100 MHz you get RT = 0.35 / 0.1 = 3.5 ns. Check your scope manual and see what you find.