> how to calculate grid resistance
Clarify the question.
The grid itself is near-infinite, at DC and bass, as long as it stays negative. Pencil in 100Megs and see if it matters.
(If it goes positive of the cathode, the grid impedance goes VERY low, like 1K, so we never do that in most audio.)
The grid resistor has to carry the grid leakage current but not load the audio source excessively. Since grid current is small, the resistor can be big. For most small tubes, 1 Meg is the maximum in all-round use. For power tubes it can be lower, expecially without self-bias, 100K or even 50K. Read the specs. In US practice, the rated max grid resistor is what we use; in EUR custom they like to go 70% of that. So for a tube rated "1 Meg Max", the US designer would use 1Meg, the EUR designer might use 680K. Not a big deal. The grid current is very variable so the max grid resistor is a worst-case spec: 99 of 100 12AX7s will be fine with 2Meg or 5Meg.
As Paul says, sometimes your source wants a certain load (instead of a super-light load). Transformers may have a suggested load for best treble response. Tone controls may use a resistor to set response. Assuming this is less than the tube's max grid resistor, you can use that to drain the grid leakage to ground.
The actual impedance looking into a grid, in audio systems, is the sum of grid leakage (~~100Megs), grid resistor (typically 1Meg), and tube capacitance. A few hundred pFd of capacitance will be down to 100K at 20KHz, 20K at 100KHz. So the total impedance runs about 1Meg from DC to 1KHz-2KHz and then falls toward 20K at 100KHz. The input impedance of a tube is NOT the grid resistor.... it may be close-enough for AM radio and 78-phono design, but in hi-fi you have to remember the capacitance. But most sources are low enough to work hi-fi: 12AX7 plate and plate resistor acts like a ~50K source, so it can drive a tube with 100pFd input capacitance up to about -3dB at 35KHz, -1dB at 17KHz, which was good-enough for many purposes.
When you get into radio carrier frequencies, the capacitance really bites. 100pFd at 1MHz is only 1.7K, and tube gain tends to about unity. We shift to low-C high-Gm tubes for wideband apps, tune-out the capacitance in narrow-band apps. But there is also a "resistive" impedance because the electrons don't move fast enough to cross in less than one cycle. RDM 3rd and 4th cover this, if you have to know. It only matters in tuned work; broadband work (and VHF tuned work) is so swamped in capacitance that a little resistance does not matter.