OpenAI debuts Voice Engine, a speech generation model that replicates the original voice

On March 29, local time, OpenAI demonstrated for the first time on its website a speech generation model called Voice Engine. The model, which is in a small preview phase, uses text input and a single 15-second audio sample to generate natural speech that closely resembles the original voice. OpenAI, which first developed the model in late 2022, is said to have used it for its text-to-speech API and for preset voices in its ChatGPT voice and read-aloud features, and said it will take a cautious and informed approach to wider distribution because of the potential risk of misuse of synthesized speech.