Meta is using artificial intelligence to translate languages better than humans

To create intelligent computers that can understand different ways of speaking, Meta came up with something really cool. They built an intelligent computer called SeamlessM4T that can understand and write nearly 100 different languages. It can even figure out what people are saying and write them down!

This cool thing called SeamlessM4T is open to everyone. They also made a special set of information called SeamlessAlign that can help computers learn better. This is a big deal for the computer world that understands speech and writing, Mehta said.

The best part is that through SeamlessM4T, people who speak different languages ​​can easily talk to each other. It's like having a magic translator ready to go whenever you need it.

This new thing from Meta reminds us of their old projects like "No Language Left Behind" and "Universal Speech Translator". These are also intelligent computers that can understand language. guess what? They use a lot of talking and writing from the Internet to teach computers. Some people didn't like the idea, they even went to court saying computer makers should give them credit or money for using their stuff.

But let's get back to the coolest thing. Meta uses a wealth of talk and writing from the Internet to teach SeamlessM4T. They made a set that combined speaking and writing, almost like a jigsaw puzzle. This helps SeamlessM4T learn how to write down what people say, translate words, and even change words from one language to another.

According to Meta, SeamlessM4T performed very well when tested. It can understand even if there is noise in the background or a different person speaking. That's because they use many different spoken and written languages ​​to teach computers, Mehta said. They are proud that the SeamlessM4T is better at this than other intelligent computers that only know to speak or write but not both.

But there are some things to consider. Sometimes these intelligent computers make mistakes. For example, they might think that doctors are always male and nurses are always female. Sometimes, if they don't know if someone is a man or a woman, they guess it's a man.

Guess you like

Origin blog.csdn.net/iCloudEnd/article/details/132471034