OpenAI Launches New Tool to Rapidly Build AI Voice Assistants

OpenAI has released a series of new tools to make it easier for developers to build applications based on its AI technology. The new real-time tools, which will be available for testing immediately, allow developers to create AI voice apps using a set of instructions.

The previous process required developers to go through at least three steps: first transcribing audio, then running a generative text model to come up with an answer to a query, and finally using a separate text-to-speech model.OpenAI derives the bulk of its revenue from companies that use its services to build their own AI applications, making the rollout of advanced features a key selling point.

Competition is also heating up as tech giants, including Google's parent company Alphabet, integrate AI models capable of processing different forms of information such as video, audio and text into their businesses.

OpenAI has introduced a post-training fine-tuning tool for models that allows developers to improve model-generated responses using images and text. This fine-tuning process can include feedback from humans, who can provide the model with examples of good and bad answers based on its responses. Fine-tuning the model using images will allow the model to have greater image understanding, enabling applications such as enhanced visual search and improved object detection in self-driving cars.

A tool that allows smaller models to learn from larger ones has also been introduced, along with Prompt Caching, a technology that cuts some development costs in half by reusing snippets of text previously processed by the AI.