Enlarge / A promotional image of Dialogue Boost. (credit: Amazon)

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.

Amazon describes how it works in a blog post:

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.

Not all content will be eligible for the dialogue boost feature, though—at least not yet. Amazon says it "has initially launched on select Amazon Originals worldwide" like The Marvelous Mrs. Maisel and The Big Sick.

While this is partly an accessibility feature for people who are hard of hearing, Amazon is also responding to a widespread complaint among viewers.

A 2022 survey found that 50 percent of 1,260 American viewers "watch content with subtitles most of the time," many of them citing "muddled audio" and saying that it's more difficult to understand dialogue in movies and TV shows than it used to be.

The dialogue problem

There's no one simple reason for this development. It involves several factors, but many of them could reasonably be categorized under a "fragmented viewer experience" banner.

Decades ago, most TVs or home theater speakers had similar audio capabilities. However, the range of devices used—from built-in speakers in cheap TVs and laptops to high-end Dolby Atmos surround-sound systems with AI-optimized sound fields and everything in between—has grown. Additionally, TV manufacturers introduce a wide variety of proprietary technologies and configurations that they can try to use to differentiate their products in marketing materials, changing the way the same content sounds on nearly every device.

All of that means that professionals who master and encode audio for distribution on streaming networks have their work cut out for them. Some shows prioritize making sure it sounds good for the highest-end setups, but the comparative lack of dynamic range in cheaper speakers can leave owners of lesser systems with muddled audio. But even if a (futile) effort is made to encode the audio for the widest viewership possible, the devices are still so fragmented that it may be impossible to ensure a quality experience for everyone.

There are other factors, too, of course. More theatrical styles of TV acting that were popular in the '70s, '80s, and '90s have given way to a subtler, more realistic delivery that was previously the realm of art films. While those deliveries played well in movie theaters with robust sound systems, they don't always work as well on $200 TVs. Streaming bitrates and related audio quality can also vary from household to household.

It's a similar set of underlying problems to TV shows appearing too dark for some viewers. Many of us remember the final-season Game of Thrones episode that had viewers squinting in vain to see what was happening. Film techniques playing on small screens with limited contrast, peak brightness, and dynamic range were all at play there—though HBO Go and HBO Now's then-subpar resolution and bitrate were a particularly big issue in that specific case.

Anyway, back to audio: Some streaming boxes or audio systems—like the Apple TV 4K or Sonos' home theater offerings—offer built-in dialogue-boosting features, but not everyone uses those devices. Amazon's new feature should, in theory, work on anything, even if it doesn't already have support for dialogue boosting.

The company hasn't announced when the feature will expand to more content. But we wouldn't be surprised to see rapid expansion—not just from Amazon, but from other streamers offering similar features, too.

Read Comments

Nicholas Weaver wrote an excellent paper on the problems of cryptocurrencies and the need to regulate the space—with all existing regulations. His conclusion:

Regulators, especially regulators in the United States, often fear accusations of stifling innovation. As such, the cryptocurrency space has grown over the past decade with very little regulatory oversight.

But fortunately for regulators, there is no actual innovation to stifle. Cryptocurrencies cannot revolutionize payments or finance, as the basic nature of all cryptocurrencies render them fundamentally unsuitable to revolutionize our financial system—which, by the way, already has decades of successful experience with digital payments and electronic money. The supposedly “decentralized” and “trustless” cryptocurrency systems, both technically and socially, fail to provide meaningful benefits to society—and indeed, necessarily also fail in their foundational claims of decentralization and trustlessness.

When regulating cryptocurrencies, the best starting point is history. Regulating various tokens is best done through the existing securities law framework, an area where the US has a near century of well-established law. It starts with regulating the issuance of new cryptocurrency tokens and related securities. This should substantially reduce the number of fraudulent offerings.

Similarly, active regulation of the cryptocurrency exchanges should offer substantial benefits, including eliminating significant consumer risk, blocking key money-laundering channels, and overall producing a far more regulated and far less manipulated market.

Finally, the stablecoins need basic regulation as money transmitters. Unless action is taken they risk becoming substantial conduits for money laundering, but requiring them to treat all users as customers should prevent this risk from developing further.

Read the whole thing.

Stanford University's IT department created a list of offensive terms and phrases accompanied by alternative recommendations. Until recently, the list was publicly available, but the university's website has made it password-protected after significant criticism.

"The Elimination of Harmful Language Initiative (EHLI) is a multi-phase, multi-year project to address harmful language in IT at Stanford," notes the list. "The goal of (EHLI) is to eliminate many forms of harmful language, including racist, violent, and biased (e.g., disability bias, ethnic bias, ethnic slurs, gender bias, implicit bias, sexual bias) language in Stanford websites and code."

The list, which was recently discovered—and ruthlessly mocked—by The Wall Street Journal will feel familiar to anyone who has ever encountered universities' microaggression reporting protocols. It includes somewhat outdated expressions with problematic racial origins that hardly anyone remembers, like "low man on the totem pole" and "long time no see." It also features many words and phrases that are inoffensive but could possibly be perceived as racial or gendered if you squinted at them long enough: "white paper," "webmaster," etc. The term "American" is disfavored on technical grounds, since not all Americans are denizens of the U.S.

Then there are a bunch of expressions to which the IT department objects because they are too vivid, including "beating a dead horse" and "take a stab at it." I would say that being against these terms is stupid or crazy, but both "stupid" and "crazy" are on the list as well.

The list also includes at least a few terms that were themselves preferred over supposedly more offensive options until very recently. A good example is "survivor," which used to appear as an acceptable substitute for "victim" but is apparently out of fashion: The IT department now prefers "person who has been impacted by." This is a good example of how attempts to make language more politically correct almost always involve making it more cumbersome.

But the list's runaway winner for most baffling inclusion is: "Karen," a term that only recently entered the cultural lexicon. Instead of saying "Karen," the IT department would like people to say "demanding or entitled White woman." (The latter strikes these ears as significantly more derisive, but I am not a Stanford guy, er, person.)

The Wall Street Journal story went viral—as entries in this category tend to—and was widely covered in conservative media. As always, one must note that neither Stanford nor other campuses are forcing students to stop saying these words. It is true, however, that dozens of schools have set up tip lines that explicitly permit students to report each other—and their teachers—for using harmful language. Unsurprisingly, Stanford is among them.

The post Stanford Seeks the 'Elimination of Harmful Language' Like 'American,' 'Stupid,' and…'Karen' appeared first on Reason.com.