Meta’s new AI speech tool helps produce public, high-quality audio clips

Meta has introduced a generative AI speech tool for users. This switch has been released under the name of Voicebox. Meta’s new AI speech tool has been brought with different features than the previous models.


Voicebox can produce high-quality audio clips in a variety of styles. By using voicebox, users can give new and fresh data for input.

Apart from this, Voicebox also allows users to modify old samples. This new AI model of Meta has been brought in support of 6 languages. Additionally, the model does separate model generation by content editing style conversations.

According to media reports, Meta’s new AI speech model has been prepared with 50,000 hours of recorded speech. In addition, the model is also trained using domain audiobooks published in different languages.

Meta’s speech model is a special model for users who cannot speak. With the model’s text-to-speech feature, it can perform speech generation.

With the text-to-speech feature, the model tries to match the audio style to the given input, after which the text-to-speech is generated. This model comes with voice customization feature for users.

The model has been brought with cross-lingual style transfer feature. With English, French, German, Spanish, Polish and Portuguese languages, users can provide speech and text input into the voice box.

Voicebox reads the given text. This feature makes communication between two people speaking different languages ​​easy and natural. With the help of this feature of the model, mispronounced words can be removed from the speech.

Images mentioned above related to are either copyright property of ICT-Samachar or respective image owners.

Rabins Sharma Lamichhane

Rabins Sharma Lamichhane is senior ICT professional who talks about #it, #cloud, #servers, #software, and #innovation. Rabins is also the first initiator of Digital Nepal. Facebook: rabinsxp Instagram: rabinsxp

You may also like...

Leave a Reply

Your email address will not be published.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