Affiliate links on Android Authority may earn us a commission. Learn more.
Bored of ChatGPT's current voices? We've got good news for you
- OpenAI appears to be working on eight new voices for ChatGPT, which would bump the total to twelve options.
- These voices will reportedly express sounds more naturally, especially when imitating animals and producing other nonverbal noises.
- The upcoming voice options will seemingly be able to convey emotions when the text they’re reading is bolded or italicized.
ChatGPT has been steadily introducing new features, consequently becoming many users’ go-to AI chatbot. One of its current perks is voice support, which enables users to converse with it or have it read text out loud. Right now, users can only pick between four voice options, which some may find a bit limiting. Fortunately, OpenAI is seemingly working to bring eight more voices to ChatGPT, raising the total to twelve.
According to Testing Catalog, recent reverse engineering reports have revealed eight new ChatGPT voices that could launch in the near future. The upcoming options — dubbed Fathom, Glimmer, Harp, Maple, Orbit, Rainbow, Reef, Ridge, and Vale — would presumably be offered alongside the existing four voices. This would increase the total to twelve.
Preview of the new ChatGPT voices (animal sounds)— Tibor Blaho (@btibor91) September 1, 2024
Beyond their distinct tones, the fresh voices will reportedly be able to express nonverbal sounds more naturally. This would allow them to imitate animals and produce other noises in a more accurate manner. The video embedded above offers some samples, including dog barks.
Similarly, the new ChatGPT voices may be able to adjust their tones based on text formatting. So, when reading bolded or italicized text, the voice may emphasize or express emotions accordingly.
It’s currently unclear when and if OpenAI will release the new ChatGPT voices to the public. For now, the fresh options are available exclusively through reverse engineering, and there’s no official way for users to access them.