Custom Mix" Project: Merging ITA vocals (from AC-3) with ENG base (TrueHD) via AI. Is this the right approach?

[pablitho]

Hi everyone,

I'm working on a custom audio mix project for my setup and I would like to get some technical feedback from those with more experience in audio restoration. The goal is to create a high-fidelity Italian track, bypassing the limitations of the classic lossy localized tracks, starting from these two sources:

1. Original ENG track in Dolby TrueHD (lossless, very high dynamic range).
2. ITALIAN dubbed track in simple AC-3 (lossy).

The workflow I'm currently using, based on iZotope RX 12 Advanced, is as follows:

• I use the Scene Rebalance module (not the classic instrumental extraction, which messes with the phase) on the English TrueHD track to lower the original vocals to -Inf.
• I extract only the Italian vocals from the AC-3 track.
• I level the ITA vocals (using Loudness Control based on the Integrated Loudness of the original vocals) and mix them over the "hollowed-out" TrueHD base, channel by channel.

At the moment, the results don't seem exceptional to me.

What do you guys think? Are there any better methods or workflows to achieve this?

Thanks in advance to anyone willing to share their experience or suggest any tips!