ChatGPT is reportedly getting 8 new voices — here’s what they sound like
OpenAI seems to be adding new voices to ChatGPT’s Advanced Voice Mode — and we might know what they’ll sound like.
New reports have discovered eight new voices that could eventually be added to ChatGPT’s current options. TestingCatalog did some reverse engineering to find the potential new voice options and engineer Tibor Blaho posted on X detailed samples of what each voice reportedly sounds like.
Check out ChatGPT’s reported new voices
The alleged new voice options are: Fathom, Glimmer, Harp, Maple, Orbit, Rainbow, Reef, Ridge, and Vale.
This would be added to ChatGPT’s existing voices, which are Ember, Juniper, Cove, and Breeze (sorry, no more Sky after that whole Scarlett Johansson debacle.) The reportedly new voices seem to vary in gender and accents, with Fathom and Vale sounding like posh Brits and Reef sounding Australian.
OpenAI announced ChatGPT Voice Mode last May as part of its new model, GPT-4o, which integrates text, vision, and audio modalities into one model. GPT-4o’s impressive capabilities were somewhat overshadowed by the demo voice’s eerie comparisons to actress Scarlett Johansson and a subsequent statement from the her actress herself saying that she turned down Sam Altman’s request to use her voice for the model.
Nevertheless, ChatGPT Voice Mode rolled out to a select number of ChatGPT Plus users in July and news about Voice Mode has stayed pretty quiet since then.
Interestingly, Blaho’s demo has the voices reading a passage that includes dog barks and crows. The voices do their best to portray the animal sounds with varying degrees of success. Some of the voices seem to be just saying “woof” and “caw,” while others attempt to actually mimic the sounds of dogs and crows. If not for these imitations, some of which sound tinny and robotic, you wouldn’t know they are AI-generated voices.
ChatGPT’s Voice Mode is quite new. So new, in fact, that the voice modality is still only available to a small amount of ChatGPT Plus users. Since it’s uncertain when the audio capability will roll out to more users, any details about ChatGPT getting more voices is also unknown.
