Cat Whisperer is right that a blunderbuss style barrel would be the best way to optimize the sound to being closer to a large gun. Think of a cannon like a horn (trumpet or even better, a compression driven loudspeaker horn).
I'm no acoustical engineer, but from what I can tell, here's how I understand it (some of this may be extraneous and/or inaccurate, but it's just my take and worth what you paid for it):
We can model a cannon for noise purposes as a compression driven horn loudspeaker.
According to horn theory, the length of a horn is key to producing the frequency desired, and must be at least 1/4 of the wavelength of the frequency to be produced, and preferably at least 1/2.
The wavelength of a sound wave is the speed of sound (343m/s) divided by the frequency. For a low frequency, say 80Hz, you'd need a length of at least 1.07m [343(m/s) / 80(1/s)]/4.
Conversely, if you want to find out the lowest frequency your length "horn" can produce, you would use the following formula:
f = c / (4*L),
where c = 343m/s, where the length is in meters,
or c = 1125ft/s, where length is in feet,
or c = 13500in/s, where length is in inches
If you have a .33m tube (~1ft.), about the lowest frequency you will be able to produce is about 258Hz. This is still a low frequency, but 258Hz will never sound like 80Hz.
Similarly, the mouth (muzzle) area is a factor in the frequency achievable according to the following formula:
A
fm = (c/F
o)
2/4pi
A
fm is the free mouth area (m
2)
c is the speed of sound (344 m/s)
F
o is the flare frequency (or cutoff frequency) in Hz
Because we are using a circular cross section, A = pi*d
2/4
This solves to:
F
o = c/(d*pi)
Using this, if we assume we have a .0127m bore (.5in), the cutoff frequency would be ~8.6kHz. As you can see if you look at the formula, the cutoff frequency is linearly inversely proportional to the bore size. If you were to increase the bore to 1in (.0254m), you would halve the cutoff frequency to 4.3kHz. A 3" bore would be capable of producing lows to ~1.43kHz.
This means that by opening up the muzzle using some form of curve - be it conical, exponential, hyperbolic, tractrix, or what have you - you can get closer to the frequency of a larger bore gun. Of course, sound pressure is just as important as frequency, which is why when you have greater pressure in the bore, you have a louder boom. That said, I wouldn't recommend a conical muzzle, at least not for the whole length, as it would remove too much material for my liking, and it wouldn't allow the wadding to expand as smoothly as the bore enlarges. I would actually recommend a combined shape of a straight bore running most of the length of the gun, say 2/3 or more, with a curve from there to the muzzle. This would allow you to use the barrel for shooting live rounds in addition to salutes. Of course, the longer the straight bore, the more accurate the gun will be when firing live rounds, while the longer the curve, the smoother and the more effective it will be in increasing the effective bore size.
Selection of wadding material would also become a factor, because that expansion I think would play a big factor in how much effect the barrel flare would impact the sound. To understand what I mean by this, consider a blunderbuss. When shooting a solid ball out of a blunderbuss, the flare has no effect on the accuracy of the gun (only your ability to sight it), while when shooting shot, it will allow the shot to spread a little more than a gun with the same primary bore size without the flare. Though there is only a small difference here, as the flare on a blunderbuss is more to serve in easing the loading of the gun, since it is a short range - often called "coach" - gun, the impact on the sound wave would be significant. I would think some powder wadding, such as flour or corn meal (perhaps mixed with a little Cremora for a nice flame) would lend itself well to this type of muzzle, as it would tend to expand smoothly throughout the curve in the muzzle.
Of course, now that I finished typing this I notice you are not even asking about a blackpowder cannon, but a toy (
http://www.bigbangcannons.com/product.aspx?i=1&c=4) that uses calcium carbide and water to, as I understand it, produce essentially trace amounts of acetylene to burn up in a flash:
http://en.wikipedia.org/wiki/Calcium_carbide. It also doesn't appear to have enough wall thickness to even try modifying in the form described above. I don't think I'd bother trying to make that thing sound more like a real cannon - you'd probably be asking for problems.