Voice interaction is becoming more and more instant and natural. OpenAI has announced the official introduction of a new "real-time dictation" feature for ChatGPT, allowing users to see transcribed text in sync while speaking into the microphone, further improving the efficiency of everyday voice input. The feature will gradually roll out to ChatGPT users worldwide, covering the web version, desktop client, and iOS and Android apps.

More intuitive operation: words appear as you speak, and you can edit them directly when you're done

Compared with the previous experience, the new feature greatly reduces waiting time. In the past, when users used ChatGPT's voice dictation, they usually had to finish recording a whole passage first and then wait for the system to convert it into text.

In the new mode, users only need to click the audio waveform button in the input box to start voice input, and they can view the converted text in real time while speaking; after recording ends, they can also edit and polish the transcribed text directly in the input box, and then send it to ChatGPT after confirming it is correct.

Removing the time cap: no more 10-minute limit, better transcription reliability

In addition to the now-synchronized interaction, this update also removes the 10-minute duration limit of the original voice dictation feature, while further improving the reliability of voice transcription, making longer dictated input smoother and more hassle-free.

In addition, ChatGPT previously introduced a "recording mode," mainly used to transcribe voice notes and meeting content, and it can organize voice information into text or bulleted key points. However, this recording mode is still in testing and is mainly available to Plus, Enterprise, Edu, Business, and Pro workspaces; by comparison, the real-time dictation feature launched this time is more widely available, further lowering the barrier for regular users to use voice input.