OpenAI particulars Voice Engine speech technology AI

OpenAI at present detailed Voice Engine, a man-made intelligence mannequin that may generate artificial speech primarily based on user-provided audio samples.

The corporate developed the mannequin in late 2022. OpenAI makes use of it to energy ChatGPT options that allow prospects to work together with the chatbot with voice instructions and have it learn textual content aloud. Moreover, the corporate made the mannequin out there to a restricted variety of companions final 12 months by way of a pilot program. 

Voice Engine can analyze a pattern of a person’s voice after which generate artificial speech that intently resembles it. In line with OpenAI, the AI requires solely 15 seconds of audio to mimic the speaker. The corporate described Voice Engine as a “small mannequin” in a weblog submit, which suggests it requires restricted computing infrastructure to run.

OpenAI has not but made Voice Engine publicly out there. Nevertheless, it opened entry to the mannequin for a restricted variety of companions in late 2023. OpenAI says these companions have efficiently utilized Voice Engine to duties resembling producing voiceovers for instructional content material and translating movies.

The corporate says the pilot program contributors agreed to copy people’ voices solely with their permission. Moreover, prospects are required so as to add disclosures to AI-generated speech specifying that it’s artificial.

OpenAI detailed that it’s taking a number of steps to make sure the pilot contributors adjust to the utilization phrases. Its engineers configured Voice Engine to include a watermark into artificial speech recordsdata. Moreover, OpenAI has launched a “proactive monitoring” initiative to make sure the AI mannequin is used responsibly.

“We hope to start out a dialogue on the accountable deployment of artificial voices, and the way society can adapt to those new capabilities,” OpenAI staffers wrote within the weblog submit detailing Voice Engine. “Based mostly on these conversations and the outcomes of those small scale checks, we are going to make a extra knowledgeable determination about whether or not and the best way to deploy this expertise at scale.”

If OpenAI decides to make Voice Engine commercially out there, the AI mannequin might create extra competitors for the present artificial speech providers in the marketplace. Eleven Labs Inc., one of many startups competing on this phase, not too long ago raised $80 million Andreessen Horowitz and different buyers. The corporate says its person base contains over 40% of the Fortune 500.

TechCrunch reported at present that Voice Engine is priced considerably decrease than Eleven Labs’ service however presents fewer customization choices. OpenAI might develop a brand new, extra superior model of Voice Engine to handle these limitations earlier than making it commercially out there. The tradeoff is {that a} extra succesful model may take extra {hardware} assets to run, which might seemingly improve its worth.

OpenAI might doubtlessly additionally open-source Voice Engine. In 2022, the 12 months it developed the mannequin, the corporate launched the code for a second AI system referred to as Whisper that may transcribe and translate speech. OpenAI detailed on the time that the latter mannequin produces 50% fewer errors than earlier neural networks within the class. 

Photograph: Unsplash