OpenAI has launched three major audio models, taking speech interaction technology to a new level.

OpenAI has launched a new generation of audio models, featuring speech-to-text and text-to-speech capabilities. gpt-4o-transcribe significantly reduces word error rates, outperforming the existing Whisper model. gpt-4o-mini-transcribe is a streamlined version that offers faster speed and higher efficiency. gpt-4o-mini-tts introduces “guidability” for the first time, allowing developers to control the voice style.