Music lovers have recently talked about this music quality problem.

I think the latter is better. The former is the quantization sampling frequency when analog signals are converted into digital signals. When it is greater than the original signal frequency, it will not be distorted when it is restored. When it is played, it will be converted into an analog signal of the earphone. Both frequencies are large enough and the difference is small. In the second case, it is generated when encoding. Of course, the greater the entropy of data, the smaller the distortion, so it mainly depends on the code rate. I use the same as you. You can't tell the difference at all.