Published in News

OpenAI opens speech-to-speech engine to developers

by on02 October 2024


Conversational voice interfaces on their way

OpenAI has given developers access to its speech-to-speech engine -- the magic behind ChatGPT's advanced voice mode.

Announced at OpenAI ‘s DevDay event, the move is expected to eventually unleash a torrent of AI apps boasting conversational voice interfaces that say, “I wouldn’t do that, Dave,” and “I’ll be back.”

Among the privileged early testers are the nutrition and fitness app Healthify and the language learning app Speak. Developers can now fine-tune models based on pictures. In a OpenAI executives showcased the new audio capabilities, seamlessly integrating with Twilio's API to have an AI assistant order 400 chocolate-covered strawberries at a fictional candy shop, which at this time of the morning just makes me hungry.

However, there's a catch. Developers are confined to using the voices provided by OpenAI, the same dulcet tones available within ChatGPT at least for now. #

These voices, unmarked and free from any identifying watermark, come with a stern warning: OpenAI's terms of service strictly prohibit using its systems to spam or mislead.

This is probably to avoid spammers using fake celebrities to hawk their products in spam robot calls.

Last modified on 02 October 2024
Rate this item
(0 votes)