Amazon introduces new feature to make dialogue in its TV shows intelligible (from Ars Technica)

[Jim Feeley]

Amazon introduces new feature to make dialogue in its TV shows intelligible

It's an accessibility feature, sure, but muddled audio is a problem for everyone.

 

SAMUEL AXON -  4/20/2023, 2:54 PM

Amazon has introduced a new feature to Prime Video called Dialogue Boost. It's intended to isolate dialogue and make it louder relative to other sounds in streaming videos on the service.

 

Rest of the article here:

https://arstechnica.com/gadgets/2023/04/amazon-introduces-new-feature-to-make-dialogue-in-its-tv-shows-intelligible/

 

 

[Olle Sjostrom]

Hmpf. We had a big debate here in Sweden over a TV series where some critics said the dialog was mumbled. I had no problem hearing the dialog at all, and it wasn't really that important. The story was told in other ways I feel, but another issue people have is: too small speakers that sit behind the TV, against a wall, in a big undampened room, with the potato chips and popcorn in your mouth, and the phone in your hand.

Basically. My parents are hard of hearing. They sit quietly in front of their iMac in a kind of dampened room. They never have an issue with audio. 

It's interesting that people claim they can't hear dialog. I am constantly amazed over how good shows and movies sound, dialog wise. And how tricky the shots are and whatnot. Anyway. AI can problably help here. But I bet the Christopher Nolans and others might have issues with Amazon using an AI to tamper with your already mixed and perfectly crafted soundtrack.

[codyman]

I can't imagine using Amazon's Fire TV interface / devices.  They are so darn slow and clunky it's not even funny.

[Jim Feeley]

59 minutes ago, codyman said:

I can't imagine using Amazon's Fire TV interface / devices.  They are so darn slow and clunky it's not even funny.

 

Ya, I've never used one of their hardware streaming devices, but my read is that this is something (that is/will be) embedded into Amazon Prime streams and not dependent on what you use to view the stream. So it's like a SAP or something. From their blog post: "Dialogue Boost is available across all devices that support the Prime Video experience." But I could be wrong.

[codyman]

34 minutes ago, Jim Feeley said:

 

Ya, I've never used one of their hardware streaming devices, but my read is that this is something (that is/will be) embedded into Amazon Prime streams and not dependent on what you use to view the stream. So it's like a SAP or something. From their blog post: "Dialogue Boost is available across all devices that support the Prime Video experience." But I could be wrong.

Ah that makes more sense.  I'm assuming they just are having substreams with the dialogue center track just normalized higher then.  Or maybe they are even telling the software to tell the Dolby decoder to decode the DialNorm metadata differently?

 

  https://en.wikipedia.org/wiki/Dialnorm

[Jim Feeley]

Don't really know what they're doing. From Amazon's blog post about Dialog Boost:

 

===========

While Dialogue Boost was built with the needs of customers who are hard of hearing in mind, anyone can use the feature to suit their personal listening preferences.

...

Dialogue Boost analyzes the original audio in a movie or series and intelligently identifies points where dialogue may be hard to hear above background music and effects. Then, speech patterns are isolated and audio is enhanced to make the dialogue clearer. This AI-based approach delivers a targeted enhancement to portions of spoken dialogue, instead of a general amplification at the center channel in a home theater system. As a result, Dialogue Boost can be enjoyed anywhere the Prime Video experience is available.

===========

 

So maybe they're doing some DialNorm-like stuff. Or maybe they integrated ChatGPT into Alexa. "Alexa, what the f...are they saying?" Maybe they really are doing some machine-learning thing (and using Christopher Nolan films as the training set). Or maybe they mentioned AI only to meet the MAISBDFU metric (Mention AI So Bezos Doesn't Fire Us).

 

An old friend and colleague works on compression for Amazon Prime. I'll drop him a note, but no idea if he'll respond...

[syncsound]

How many channels will it apply to? If it's center-only, then dialogue that pans may undergo a noticeable change, one that detracts. Granted, actively panned dialogue isn't common to that many titles (Gravity and Roma  are the first that come to mind).

[Derek H]

As misguided as this is in some ways I’d argue it’s a better solution than just accepting defeat and letting people default to watching shows with subtitles on.
 

You can’t fault Amazon for trying to do something about this problem.  The Audio blame game around this issue is ridiculous and if no one is willing to just ask for more intelligible dialog out of the source and prioritize better, quieter, sets and favorable camera blocking well then…

 

Looking forward to the day when an impatient AD tells me “don’t worry about it the AI will fix it!”

[fuzztech]

I am with Derek. It seems that the emphasis has been moving away for some time from great audio to loud audio and tons of background sounds, which can be great until it isn't. It strikes me that if one needs special processing to hear the main characters speaking, something is wrong with the production values. Yes, I'm old. 

[codyman]

1 hour ago, fuzztech said:

It seems that the emphasis has been moving away for some time from great audio to loud audio and tons of background sounds, which can be great until it isn't.

I think a lot of this is to build rich environments around a bunch of lav dialogue which usually is sterile and noise reduced.  With 3+ cams doing "wide and tight", there isn't a lot of boom playing these days it seems.  Don't get me wrong, iZotope RX is a fantastic tool and performs magic, but often times if you clean up a lav, it's kind of in a vacuum almost.

[fuzztech]

Hmm. Surely most things we will get in a final feature on Amazon are going to be ADR, right? I've been out of the game for a while but if we're talking about the kind of production that would need this help, I'd reckon (I am asking a question, not making a statement) that lav-audio would've been replaced in the ADR world? 

[codyman]

Just now, fuzztech said:

Hmm. Surely most things we will get in a final feature on Amazon are going to be ADR, right? I've been out of the game for a while but if we're talking about the kind of production that would need this help, I'd reckon (I am asking a question, not making a statement) that lav-audio would've been replaced in the ADR world? 

I haven't done post on a show of that scale, but I find at least in television, it seems to be a mish mash of original and ADR (at least to my ears).  I'm pretty sure all your big blockbusters of course are dubbed, especially all the big Marvel action stuff.  ADR matching though has come to be really good overall these days though.  We've come a long ways on that!

[fuzztech]

Right on!