OpenAI has applied for a trademark called 'voice engine' to construct a synthetic intelligence (AI) voice assistant. It is anticipated that voice assistants equivalent to Apple's 'Siri' or Amazon's 'Alexa' shall be released ahead of 'GPT-5', which is anticipated to seem in the course of this 12 months.
Business Insider reported on the twenty third (local time) that OpenAI submitted a trademark application to the USA Patent and Trademark Office (USPTO) on the nineteenth. “The Voice Engine trademark registration is meant to cover the categories of downloadable computer programs, downloadable computer software development tools, and downloadable computer software for speech and voice recognition, voice command processing, and conversion between text and speech,” the document states. It's done.
This trademark application was filed the day after OpenAI CEO Sam Altman said in a podcast interview with computer scientist Rex Friedman, “I even have something necessary to disclose ahead of GPT-5.”
“We're launching an incredible latest model this 12 months. I don't know what we're going to call it, but we're launching a wide range of products in the following few months. It's going to be really cool,” Altman said on the printed.
“The performance of GPT-5 may meet or exceed expectations,” he said. “But we must disclose other necessary details first.”
In other words, the evaluation is that a latest model shall be released prior to the launch of GPT-5, and that it’s going to likely be in the shape of a voice assistant as a consequence of the patent application. Along with voice assistants, there may be a possibility of releasing other models, equivalent to music creation AI or other models that OpenAI has never released.
OpenAI already has a TTS API that converts voice into text, and has also released an open source voice recognition model called 'Whisper'. Nevertheless, the Voice Engineer trademark application submitted to the USPTO this time includes software in 10 fields.
▲Constructing a digital voice assistant ▲Voice and voice recognition, voice command processing, text-to-speech conversion ▲Voice command processing and text-to-speech conversion ▲Automatic speech recognition and generation ▲Natural language prompts, based on text, voice, visual prompts, images and videos ▲ Generate speech and audio output in response to user prompts ▲ Machine learning-based natural language and speech processing ▲ Multilingual speech recognition, translation, and transcription ▲ Use AI for automatic text-to-speech and text-to-audio conversion ▲ Speech services Delivery development ▲API ▲SDK, etc.
Based on this, it seems that a voice assistant shall be created that understands the user's voice, text, images, videos, etc. and responds with voice or sound. Foreign language translations are also included.
In fact, it’s unclear whether trademark registration shall be successful. An organization called Applied Voice Technology also applied for the trademark Voice Engine in 1989, however the registration was not accepted.
Prior to this, last month, the USPTO also rejected OpenAI's application for the 'GPT' trademark. GPT stands for ‘Generative Pre-trained Transformer’, and it’s because it is solely a term that describes the characteristics of a product and is nothing greater than a widely used abbreviation.
Nevertheless, trademark registration for 'GPT-4' has been permitted, and screening for 'GPT-5', 'GPT-6', 'GPT-7', 'Sora', and 'GPT Store' is in progress.
Reporter Park Chan cpark@aitimes.com