Polyphonic Challenge Part 2

Now I see polyphonic ciphers do not pose a serious problem to codebreaking. Part 2 of my "Polyphonic Challenge" has been released at MTC3 (Part 1 was reported here), and three people already solved it.
Part 1 and Part 2 have different characteristics. While the digits of Part 1 tend to represent two letters of similar frequencies, the digits of Part 2 tend to represent one high-frequency letter and one low-frequency letter. I think Part 2 is easier. Although frequencies of digits are somewhat flattened, considering a reduced alphabet consisting of only high-frequency letters may reveal some portions of the plaintext. (Anyway, I have not done it myself, and I hope someone will demonstrate solution of a ciphertext in a polyphonic substitution cipher.)