Newbie question: Hybrid audio restoration using English Lossless M&E + Local Lossy Di

[pablitho]

Hi everyone,
I’ve recently started getting into the world of audio preservation and restoration, and after doing some research, I stumbled upon this forum. I have to say, the level of technical expertise here is absolutely amazing.
I was thinking about a specific approach to hybrid audio restoration, and I wanted to ask if this is something you have already experimented with, or if it's considered a standard practice here.
When restoring a local release (in my case, Italian), we often face the issue of great older mixes being trapped in lossy formats (like AC3 or standard DTS from old DVDs), while the modern Blu-ray/4K releases have an incredible, dynamic Lossless English track (DTS-HD MA or Dolby TrueHD/Atmos).
Has anyone ever tried to build a "hybrid" lossless track by keeping the untouched English Lossless channels for everything related to M&E (Music & Effects) — meaning Left, Right, Surrounds, and LFE — and then extracting and sync-matching only the local dialogue from the lossy center channel onto the lossless English center?
Conceptually, this seems like the ultimate way to get the best of both worlds: the full dynamic range of the original lossless effects/score, combined with our local dubbing.
However, being new to this, I can foresee a few technical roadblocks, and I'd love to hear your thoughts on them:
1. Bleeding: How do you deal with dialogue bleeding into the Left and Right channels on older local mixes?
2. Center Channel M&E: Since the English center channel also contains specific localized sound effects and foley, replacing it entirely with the lossy local center might degrade the ambient quality. Do you use spectral isolation (like iZotope RX) to extract only the dry vocal track from the lossy source before mixing it into the English track?
3. Phase & Mix Differences: How painful is it to level-match and EQ the hybrid track when the original theatrical audio stems for the two languages come from completely different masters?
I would love to hear your experiences, tips, or if there are any specific threads/guides covering this kind of "vocal transplant" workflow.
Thanks a lot for your time and for this incredible community!

[stwd4nder2]

I mean you could, but tbh I don't think it's worth it. Often the modern releases, while lossless, suffer other issues such as the mix being altered or EQ'd to remove detail and dynamics. So the lossy track tends to be overall better than the lossless. And even if that wasn't the case, lossy ac3 is honestly good enough for most cases (it was used in theatrical prints after all).

[pablitho]

That’s a very fair point regarding revisionist mixes, and I completely agree that an untouched older mix is often structurally superior to a compressed modern one.

However, my question was aimed precisely at those cases where the English lossless track is undeniably, vastly superior in terms of pure dynamic range, bass extension, and detail, while the local Italian track is stuck with a standard, flat lossy encode.

Take films like The Dark Knight, The Dark Knight Rises, or Michael Mann's Collateral. On these titles, the English lossless track is a powerhouse—the gunfights in Collateral or the IMAX sequences in Nolan's films have an aggressive, visceral sound design that completely destroys the flat, constrained Italian lossy tracks we got on the retail discs.

In cases like these, where the English lossless mix is clearly the definitive way to experience the film's sound design, is a hybrid project technically viable? >
If I wanted to extract the Italian dialogue from the lossy track and "graft" it onto the English lossless M&E base for these specific films, how would you approach it practically? Is it a nightmare of phase issues, or is it doable with tools like iZotope RX for spectral separation?

I’d love to know if anyone here has ever attempted this kind of "vocal transplant" on action-heavy movies like these.

[pablitho]

Following up on my previous post! As I mentioned, for movies like The Dark Knight, The Dark Knight Rises, or Collateral, the English lossless track is an absolute powerhouse. The visceral sound design completely destroys the flat Italian lossy tracks, which is why I really want to pull off this hybrid "vocal transplant".

Since my last post, I've been brainstorming a potential technical pipeline with an AI (Gemini), and it suggested a highly specific, "state-of-the-art DSP" workflow to solve the exact roadblocks I mentioned earlier (L/R bleeding, center channel degradation, and phase nightmares). I'd love to get your reality check on this AI-generated pipeline to see if it makes sense in the real world.

Here is the proposed workflow:

1. Center Channel Extraction & Bleed Removal
Instead of relying solely on iZotope RX (which can leave watery artifacts when pushed hard), the AI suggested using Steinberg SpectraLayers Pro. To fix the issue of M&E bleeding into the extracted local dialogue, it recommended the 'Cast' (Imprint) process. By using the isolated instrumental stem as a negative mold, SpectraLayers can use the spectral amplitudes to literally carve out the residual music bleed from the vocal track.

2. Upscaling the Lossy Local Dialogue
Since the local Italian track is stuck with a standard lossy encode, it lacks high-end detail. The plan is to use tools like iZotope Spectral Recovery or Accentize dxRevive Pro. These tools analyze the band-limited audio and resynthesize the missing upper frequencies (above 4kHz) to give the lossy voice the same presence and frequency range as the lossless environment.

3. Timing and Phase Alignment (The Phase Nightmare)
To fit the local dialogue into the English track, the AI suggested using Synchro Arts VocAlign Ultra to automatically warp the timing and pitch of the Italian dub to match the English guide track. Then, to avoid phase cancellation when blending this dry vocal with the lossless center channel's native Foley and room tone, the suggestion is to use Sound Radix Auto-Align Post 2 to dynamically fix phase shifts.

4. Phantom Center & L/R Dialogue Bleeding
For dialogue that bleeds or is panned into the Left and Right channels, inserting a dry mono vocal ruins the ambient quality and soundstage. The AI recommended using Accentize Chameleon Surround to extract a 3D impulse response from the original English phantom voice, and wrap that exact acoustic space around the dry Italian vocal. For this kind of matching, a convolution reverb is essential to simulate the acoustic properties accurately so the ADR doesn't sound out of place.

5. Center Channel "Airfill"
To mask the edits in the center channel (where the English dialogue was removed) without destroying the underlying Foley, the plan is to isolate a tiny, clean seed of the original room tone. Then, use a convolution reverb (like Altiverb) and run white noise through it at a very low level to synthesize an endless, non-looping room tone to fill the gaps.

What do you guys think? Has anyone successfully used SpectraLayers' Cast feature for this kind of extraction? Is this AI-generated pipeline a viable solution to preserve the dynamic range of Collateral and The Dark Knight, or am I still walking into a phase nightmare?

Thanks again for your input!