Audio: EQ-IIR: add Direct-1 format implementation

[andrula-song]

Add C version Direct-1 format implementation for IIR. Compared with the original Direct-2 transport format, Direct-1 has better performance on low frequency since lower quantization noise.

Signed-off-by: Andrula Song [email protected]

[andrula-song]

Tested by xtensa simulator, the cycles that C version of Direct-1 cost is almost the same with Direct-2 transport.

[andrula-song]

here are the detailed test results, DF1 and DF2T have almost the same performance: and the cycles cost for C version are almost the same, but DF1 can reduce about 14.5% than DF2T on HiFi 3 version:

[singalsu]

here are the detailed test results, DF1 and DF2T have almost the same performance: and the cycles cost for C version are almost the same, but DF1 can reduce about 14.5% than DF2T on HiFi 3 version:

Good result, not much difference! Actually for both DFs using a bit less conservative biquads scaling in coefficients export could improve THD+N. It's some possible work for future.

Also the response comparison vs. decoded blob could compute a RMS error value as improvement. But I'd expect DF1 to be at least as good as 2T.