Voice Cloning Redefined: OpenAI’s Model Breakthrough In Just 15 Seconds

The model has been developed since late 2022 and powers the Read Out loud component in ChatGPT.

OpenAI is offering restricted admittance to a text-to-voice age stage it created called Voice engine,

which can manufactured voice in light of a 15-second clasp of somebody’s voice.

The artificial intelligence produced voice can peruse out text prompts on order in similar language as the

speaker or in various different dialects.

“These limited scale arrangements are assisting with illuminating our methodology, protects, and contemplating

how Voice Motor could be utilized for good across different businesses,” OpenAI said in its blog entry.

Organizations with access include:

Schooling technology organization “Period of Learning”

Visual storytelling platform “HeyGen”

Advanced health software producer “Dimagi”

Artificial intelligence communication application creator “Livox”

Health system “Life expectancy”

In examples provided by OpenAI, you can hear how “Age of Learning” has been using the technology to create

pre-recorded voice-over content, as well as reading out “real-time, personalized responses” to students generated by GPT-4.

OpenAI mentioned that they began developing Voice Engine in late 2022. This technology has previously powered pre-recorded voices for the text-to-speech API and ChatGPT’s “Read Aloud” feature.

In an interview with TechCrunch,

Jeff Harris, a member of OpenAI’s product team for Voice Engine, stated that the model was trained on

“a combination of licensed and readily available data. “OpenAI told the distribution of the model may be accessible to about 10 designers.

AI text-to-speech generation is an area of artificial intelligence that is advancing.

While most focus on instrumental or natural sounds, fewer have concentrated on voice generation, partly due to the concerns OpenAI mentioned.

Some notable companies in this field include Podcastle and ElevenLabs, which offer AI voice cloning technology and tools, as explored by The Vergecast the previous year.

The US government is taking steps to regulate unethical uses of AI voice technology. Recently, the Federal Communications Commission banned robocalls using AI voices after people received spam calls featuring an AI-cloned voice of President Joe Biden.

According to OpenAI, its partners have agreed to adhere to usage policies. They won’t use Voice Generation to impersonate individuals or organizations without their consent.

They also must get the “explicit and informed consent” of the original speaker, refrain from enabling individual users to create their own voices, and disclose to listeners that the voices are AI-generated. OpenAI has also added watermarking to the audio snippets to track their origin and track their usage.

OpenAI has recommended several steps to reduce

the risks associated with tools like these:

Decreasing reliance on voice-based authentication: This means finding other ways to secure accounts instead of relying solely on voice recognition, which could help prevent AI-generated voices from being misused for fraudulent purposes.

Establishing policies to protect people’s voices in AI: Creating clear rules and regulations to prevent unauthorized use or alteration of individuals’ voices in AI applications.

Increasing education about AI deepfakes: Educating the public about the existence and potential dangers of AI-generated deepfake content, and helping people learn how to identify and deal with them.

Improving AI content moderation systems: Enhancing the capabilities of AI-driven systems to detect and flag AI-generated content, which can help reduce the spread of misinformation and harmful deepfakes.

These steps aimed at addressing concerns about the misuse of AI-generated voices and promoting responsible development and use of AI technology.