Meta has updated the SeamlessM4T AI translator, making it smoother and more expressive

Meta has unveiled the second version of its SeamlessM4T multimodal neural network for speech translation. The update makes interpreting more spontaneous and emotional.

Here's What We Know

The first new feature, SeamlessExpressive, brings the intonations of the original audio into the translation: volume, pitch, tempo, pauses, etc. This gives the conversation a more natural feel.

The second feature, SeamlessStreaming, starts the translation while the person is still talking. This reduces the delay to two seconds and does not have to wait for the interlocutor to finish the phrase.

According to Meta, the algorithm analyses the part of the sentence that has already been spoken and decides if there is enough context to start translation.

The company has not yet given an exact timeline for when the new features will become available to a wider audience.

Go Deeper:

• Meta has launched the Seamless M4T, an AI-powered model that translates text and speech into 100 languages

Source: Meta