Meta back with a new banger and set the bar on the next stage of translation tools. Recently the brand has released its new SeamlessM4T multilingual multimodal translation model, which offers the facility of speech and text translation in up to 100 different languages.
Meta has announced that its SeamlessM4T model is eligible to understand both voices as well text inputs and can translate in speech-to-speech, speech-to-text, text0to-speech, and text-to-text translation as well transcription for up to 100 languages, all of these are depending on the task.
This SeamlessM4T multilingual multimodal translation model helps the users to communicate effortlessly and fluently without any misunderstanding through voice and text across different languages.
SeamlessM4T supports
- It helps in speech recognition in approx 100 languages.
- For Speech-to-text translation, it offers nearly 100 input and output languages.
- For Speech-to-speech translation, it offers about 100 input languages and 36 output languages, including English.
- It also supports 100 languages for Text-to-text translation.
- Text-to-text translation offers 35 output languages, including English and 100 input languages.
The brand is also announcing the metadata of SeamlessAlign, the hugest open multimodal translation dataset to date, with overall 270,000 hours of mined speech and text alignments. This SeamlessM4T single system proceeds towards reducing errors and delays. Apart from this, it also helps increase the translation procedure’s logicality and quality.
SeamlessM4T authorised the people who used to speak different languages to communicate with each other more impressively. Meta and other brands have already been working on creating a universal translator for the last few years. For more info, you can click here;
“If you like this article follow us on Google News, Facebook, Telegram, and Twitter. We will keep bringing you such articles.”
