ChatGPT has learned to transcribe speech into text on the fly without time limits

Источник изображения: OpenAI (изменено)

OpenAI has improved ChatGPT’s voice input feature by adding real-time speech transcript display. At the same time, the company lifted the limit on the duration of voice recordings of 10 minutes and announced an increase in the accuracy of the function.

  Image source: OpenAI (modified)

Image source: OpenAI (modified)

While recording, simply click on the sound wave icon to open the text that ChatGPT currently recognizes. Once the dictation is complete, the resulting transcript can be edited directly in the query input field before sending it to the AI.

OpenAI says the update not only makes speech recognition faster, but also makes it more reliable. At the same time, the changes affect the function of dictating voice requests, which is not a full-fledged ChatGPT voice mode that allows you to have a two-way conversation with AI.

Previously, the company also introduced a separate Record mode for transcribing voice notes and meetings. However, it remains a beta version for now and is only available to users of Plus, Enterprise, Edu, Business and Pro tariffs.

The new feature is rolling out globally across all ChatGPT plans and is available from November 19 across the web, desktop, iOS and Android apps.

If you notice an error, select it with the mouse and press CTRL+ENTER.