Designing Pro Logic II–style upmix on ADAU1701 (SigmaStudio)

Hello sreeharikrishna,

I do not have an example project handy that exactly does what you are asking about. Perhaps another user on the forum might chime in!

What I will attach are several projects that you can take some inspiration from and create what you are looking for.

Let me start with this one:

As you see by the project name, it extracts the phantom center content. So anything that is exactly in phase on the left and the right will be pulled out and be available as its own channel. However, this is a simple project and you could take it further and try to subtract the center content from the left and the right.

1261.ADAU1701 Phantom Center Channel Extracting.dspproj

This next one is a crossover where it is used to extract a subwoofer signal. This could easily be modified to produce an LFE channel.

0257.ADAU1701 Crossover with Sub Output.dspproj

This last one is more interesting to you. This one is done on the ADAU1452 DSP but you can easily copy what is done here and do it on the ADAU1701.

There are four different ways to do spatial widening in this project.

1) using the Hilbert Transform

2) using the L-R and R-L method

3) Using both the Hilbert Transform and the L-R R-L methods.

4) Using the Phat Stereo block

Start experimenting and have fun!! This is the great part about SigmaDSP. It is easy to experiment!

Here is where you can find these blocks:

Dave T

Hello Dave,

Thank you for sharing the reference projects — I’ve gone through the phantom center extractor and the crossover examples.

I would like to clarify my actual expectation, as my requirement goes significantly beyond simple L+R center extraction or basic L−R surround.

What I am aiming for is a Pro Logic II–style adaptive upmixer behaviour, specifically:

Center channel

Not just (L+R)/2

Center should be strongly voice-focused

Instruments (even mid-band instruments) should be heavily suppressed

Center extraction must be adaptive, not static — it should behave consistently across:

high-quality movie sources

compressed / low-quality stereo

streaming / Bluetooth sources

Front Left / Right

Front L/R should be instrument-rich, with minimal to near-zero vocal leakage

When dialog is present, vocals should be strongly steered to center, not merely reduced by gain subtraction

Instruments that overlap the vocal band should remain in front channels, not disappear

This implies dynamic steering, not fixed L−α·Center math

Surround

Basic L−R works, but I am looking for a cleaner surround extraction

Needs noise/crackling control when dialog is present

Should avoid vocal modulation artifacts with unbalanced stereo sources

I understand exact Dolby logic is proprietary — I’m not looking for licensed algorithms, only architectural guidance on how to build a robust, adaptive block structure on ADAU1701 that behaves similarly in real-world content.

Expecting a high advanced front left and right channel extraction