AI-Powered Closed Captions Could Open Up New Possibilities – and Pitfalls

[ad_1]

Closed headlines have become a staple of the experience of watching TV and film. It is a way for some Decryment dialogue dialogue. For others, like those who are deaf or heard, it is critical A tool. But the headlines are not perfect and technological companies and studios are increasingly looking at AI to change it.

The title for television shows and films is carried out by real people that can help provide accuracy and maintain the nuance by real people. But there are difficulties. Anyone watching a live event with closed posts disappear in the text often disappears in the text and there may be errors in a hurry of the process. The scripted programming offers more time for accuracy and detail, but it can still be a labor-intensive process – or may cost the expensive in the eyes of studios.

AI Atlas engagement tag

Warner Bros. in September. Discovery, went to the team with Google Cloud Develop AI-Free Closed Posts“combined with human control for quality assurance.” When a press release, the company reduced the title of up to 80% using the AI ​​discount using a discount to 80%. Experts say this is a peek to the future.

“Anyone who does not do it is only waiting to be displaced,” Joe Devon, web accessibility and co-founder Global Accessibility Consciousness DayHe said to use the EU in the title. The quality of the headlines with today’s hand is “everywhere, and must be developed.”

As the AI ​​continues to change our world, it’s also How to practice the acquisition of companies. Google’s”s Expressive heads The feature uses AI to better deliver emotion and tone in videos. Added apple transcription As the ways of being more accessible to voice messages and memories in iOS 18. Both Google and Apple make rich Title tools in real time to help people or hearing people Log in to the audio content added on their devices and Amazon Text-to-speech and title features Alexa.

With a computer screen and an international stage of home hunters with AI title program, with an international stage of home hunters, with the read title: "[narrator]    Lisa leaves to chase a dream from Boston and his work"

Warner Bros. Discovery combines with Google Cloud to carry AI-pighte weather. Controls a human process.

Google / Warner Bros. Discovery

In the entertainment space, Amazon began a feature in 2023 Increasing dialogue Background music and effects are in the main video that uses AI to identify and develop speech that can be difficult to hear above. The company announced the pilot program using AI in the EU in March in the EU “Otherwise, there would be no other dubbet Blog Post. Also, the gathered confident watchers were in the headline and spread to Netflix in April Dialogue only the subtitle selection For anyone who wants to understand what it means to say what the conversations mean to be in conversations.

As the EU continues to grow and consumes more content on both large and smaller screens, more studios, networks and technological companies have entered the potential of the AI ​​- I hope that the closed headlines are primarily.

To get in the foreground

Started as to develop a closed title in the United States In the 1970s, the eventAs a result, everything from live television is more equally equal to the film blockbusters for a wider audience. However, many observations of deaf or hard, also prefer to watch the films and television shows – this is due to the translation of the language, especially in cases of production dialogue hard to decipher.

Half of Americans According to the 2024 survey prepared by the language learning area, 55% of respondents have become difficult to hear dialogue in the film and shows. These habits are not limited to older viewers; 2023 found the Yougov Survey 63% of adults under 30 They prefer to watch TV with 30% of people over 65 years old.

“People and also the creators of content are inclined to assume captives, only for the toughs of the deaf or hearing society,” said Ariel Simms, President and CEO Disability belongs to. But the headlines can facilitate everyone to process and store information.

Accelerating the title process, you can get more content whether AI, TV show, movie or social media clipes, simms notes. However, the quality can suffer, especially in the first days.

“We have a name for the headlines arising in the disability community – we call them ‘executionerTions, ‘”Simms laughed.

These automated articles still fight things like punctuation, grammar and correct names. Technology will not be able to get different emphasis, dialects or speech patterns in the way a person wants.

Ideally, Simms will still have a person to maintain the accuracy and quality of companies using the AI ​​to prepare AI. Studies and networks should also work with a direct disability community to avoid compromising in the process.

“I’m not sure that we can easily remove people from the process.” “I think the technology will continue to improve. However, we were not partner with the disability community at the end of the day, we are out of an incredibly important prospect for all these possible instruments.”

Warner Bros. Studies like Discovery and Amazon, for example, it is clear to emphasize the role of people in providing a person’s energetic title and dubbing.

“If you allow your EI Slop to dominate your content, you will lose your reputation,” Devon said. “This is where the person will be in the loop.”

However, given how fast the technology is developing, the participation of man’s participation can last forever.

“Studios and broadcasters will do everything that ends the least,” he said. However, he added: “If the technology gives a power to make the work better, who is someone in this way?”

The line between detail and extreme

This is not only TV and movies where the AI ​​is a super boot title. Social media platforms such as Tiktok and Instagram Performed automatic title features to get more content.

These native titles often look like a plain text, but sometimes the creators choose flashier screens in the editing process. When using a common “karaoke” style, different colors for text, it emphasizes every individual word, as he speaks. However, this is a more dynamic approach, while the eye can compromise. People can’t read at their pace and can distract all the colors and movements.

“There is no way to satisfy 100% of users, but only a small percentage benefit from the karaoke style,” he said Meryl K. EvansAccessory marketing consultant with deaf. He says it should watch a video with dynamic writings several times to receive a message. “The most accessible articles are bored. They allowed the video to be a star.”

But there are ways to protect simplicity when adding a useful context. Google’s expressive writing features use the EU to emphasize certain voices and give a better idea of ​​what happens on the phones of viewers. An exciting “Holy Birth!” For example, all the lids may appear, for example, or an enthusiasm of a sports kicker, “Amaaaazing Shot!” Expressive articles also sound like stickers applause, drilling and fit. The text on the entire screen appears in black and white, so it does not distract.

Expressive topics used during a football game describing some words on all lids.

Examable topics have everything in some words to excite.

Google

The main attention was paid when developing the opportunity, but Android’s product management director Angana Ghosh said that the team will benefit from the use of the use of deaf or hearing. (Think about what you are in the people without a hoodless people, but still want to follow what happens in a video.)

“When we develop for accessibility, we really build a better product for everyone,” Ghosh says.

Again, some people can prefer more live recordings. In April, the name agency fcb debuted a platform with AI-power called Chicago Title with intentUsing animation, color and variable types to deliver emotion, tons and sliding. Different text colors represent different character lines and the words are emphasized and synchronized in the actor’s speech. The variable type of sizes and weight helps someone talk, as well as how their intonation is. Open source platform are available for studios, production companies and flow platforms to implement.

The FCB Chicago exploded to develop and test the change of title with deaf and hearing people with the Hearing Society. Bruno Mazzotti, Executive Director of CEO in FCB Chicago, said that the two black parents helped their experience form a platform.

“The closed title was a lot of my life; it was a factor that decided to follow our family,” Mazzotti said. “I have a hearing privilege, when the texts are not always in the work, when the headlines are in the dialogue, or more than one person suddenly, the main goal was to bring more emotion, tons and dynamic identities to people.”

A scene from the forrest GUMP, which is a title that reads, "What is your only goal in this army?"

The intention is a platform that uses animation, color and different types to deliver hoods, tones, emotion and walking.

Title with intent

Finally, Mazzotti is to present more customization options, so the goal is to offer more customization options to regulate the title intensity of viewers. Again, more animated approaches can be very distracted for some viewers and can make it difficult to watch what happens on the screen. As a result, it boils on the individual choice.

“It is not to say that we have to decide such approaches,” he said. “But we need to learn to carefully to ensure that snow and hearing followers have a net benefit.”

No Easy Edit

Despite the existing shortcomings, AI can eventually help expand the existence of headlines and offer greater customization.

YouTube Automatic headlines How, how, Despite a rough startAI, especially the technology can get more video content as it improves time. It may be a future that the topics are adjusted to different reading levels and speeds. Data without speech can be described more, thus, you will learn more to deliver the mood instead of general tags such as “scary music”.

But the learning curve is upright.

“AI headlines, especially the best of audio knowers, especially the sound quality, both television and film are widespread,” he said. Hallucinations can also serve inexact articles that areolate the ear and hard hearing followers. Therefore, people must remain part of the heading process.

It is likely that the work will happen, Deborah Bakers, Director of the Inclusive Media and Design Center of Toronto Metropolitan University, Deborah Wellas, Deborah Schelas and Design Center, he said. Man’s heads will predict that the AI ​​will predict that once it will control fifty labor.

“So now, we have a different job in the title,” he said. “People are better in finding the mistakes and decide how you make them fix them.”

For the title, the EU is still a naughty technology that is limited to a handful of companies, it will probably not be such a long time.

“They all go in this direction,” he said. “This is a matter of time – and not so much.”



[ad_2]

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *