Vai al contenuto
AI.info

Model Tracker

Voicebox

Voicebox is an AI model made by Meta. Meta released it on 16 June 2023. It takes text, audio and returns audio.

In inglese

Meta

Released
16 June 2023 — ai.meta.com, read 16 October 2026
Status
Not confirmed: The reviewed Meta pages announce Voicebox as research but do not state a lifecycle status matching the register's status values.
Takes
Text, Audio — ai.meta.com, read 16 October 2026
Returns
Audio — ai.meta.com, read 16 October 2026
Open weights
No — ai.meta.com, read 16 October 2026
Licence
Not confirmed: The reviewed Meta pages do not name a licence for Voicebox.
Context window
Not confirmed: The reviewed Meta pages do not print a context-window limit for Voicebox.
Longest answer
Not confirmed: The reviewed Meta pages do not print a maximum output-token limit for Voicebox.
Knowledge cutoff
Not confirmed: The reviewed Meta pages do not state a knowledge cutoff for Voicebox.
API price, input
Not confirmed: The reviewed Meta pages do not provide a standard API input price for Voicebox.
API price, output
Not confirmed: The reviewed Meta pages do not provide a standard API output price for Voicebox.
Where
Not confirmed: the reading gave a value this register cannot hold
Parameters
Not confirmed: The reviewed Meta pages do not publish a parameter count for Voicebox.
Model card
Not confirmed: The reviewed Meta pages link to a research paper, but do not identify or provide a Voicebox model card.

Questions

When was Voicebox released?

Meta released Voicebox on 16 June 2023.

Are the weights of Voicebox open?

No: Meta does not publish the weights of Voicebox.

What does Voicebox take and return?

Voicebox takes text, audio and returns audio.

More from Meta

Sources

  1. Introducing Voicebox: The first generative AI model for speech to generalize across tasks with state-of-the-art performance (read 16 October 2026)
  2. Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale (read 16 October 2026)

All of the Model Tracker