Open AI has upgraded the high -end voice function (AVM) of ‘Chat GPT’. In response to the recent trend, the intention of creating AI chatbots feel more human, and a few problems have been revealed.
Open AI announced on the seventh that it has launched AVM function update on X (Twitter) and its homepage. From today, it was first applied to paid users.
The intonation and naturalness of the voice are introduced to greatly improve and interactions are more natural and natural as humans. “It provides more accurate expressions of specific emotions corresponding to more natural voice, delicate accent, realistic rhymes (including stops and emphasis), and empathy and sarcasive words.”
The interpretation function has also been improved. For those who just ask for a translation between languages, you’ll proceed to translate throughout the conversation until you’re ordered to stop or turn.
Earlier this 12 months, Open AI has released AVM to free users earlier this 12 months. As well as, updates corresponding to reducing the act of breaking users and improving accents.
As well as, this update was made almost concurrently the Eleven Labs released the voice model ‘Eleven V3 Alpha’. As such, the function of ‘AI voice that feels like humans’ is emphasized.
Nevertheless, this update has been raised through communities corresponding to Reddit.
Open AI said, “This update can result in some decrease in sound quality, and unexpected tone and negative changes can occur. We expect that audio consistency will improve in the longer term.”
“In voice mode, hallucinations that occur rarely continued, leading to unintentional sounds much like promoting, gibberish, and background music,” he said. “We’re actively investigating the issue and finding solutions.”
By Dae -jun Lim, reporter ydj@aitimes.com
