Generates an article from an audio file

Imagen de ficheros de audio convertidos a texto
Imagen de ficheros de audio convertidos a texto

The system included in the free Cibeles AI plugin, incorporates a new button whose functionality allows the editor to upload an audio file with mp3, mp4, mpeg, mpga, m4a, wav or webm extension and generate a literal transcription of text, thanks to OpenAI’s Whisper model.

In addition, the interface allows you to edit the transcription, in case you want to start from it, or if you want to make any correction, such as a proper name or missing information in the audio or interview. That is, if we already have the text transcript instead of the audio, we can paste it directly into the box and generate the article, otherwise, we upload the audio and the transcript will be generated.

Based on the transcript and an elaborate Prompt, the system will generate a journalistic-style article, with almost no hallucinations and high accuracy, using the GPT4-turbo model.

We can click on the “Generate item” button as many times as we want, returning a different one each time. We can also click on the “Generate transcript” button as many times as we want, but since it is literal, it will always return the same text.


In short, the idea is to give journalists more power over the production of information without losing control, by integrating services they frequently use into their daily toolkit.

Scroll to Top