The impact of whisper on journalism and translation: The AI innovation that is revolutionizing audio transcription

2 minutes read

Discover Whisper, the artificial intelligence technology developed by OpenAI that is revolutionizing audio transcription. With its ability to transcribe audio files with exceptional accuracy, Whisper is becoming an essential tool in various fields such as journalism and translation.

Table of Contents

What is Whisper?

Whisper is an artificial intelligence-based technology for transcribing audio files. Unlike common free tools that often have errors such as word mix-ups, misplacements, or the inclusion of made-up data, Whisper offers a reliable and effective solution. Simply upload an audio file to its system, which then analyzes it and transcribes all the words spoken in the audio. OpenAI offers Whisper as a much more reliable tool for transcriptions.

Whisper, in its current version, is an automatic speech recognition (ASR) system, using AI to process audio files and convert them to text. This version was trained with over a million hours of audio, surpassing its previous version’s 680,000 hours and reducing errors by 10 to 20 percent.

Whisper’s performance

Currently, Whisper has an error rate of less than 5% when transcribing into Spanish, making it one of the best tools. It can also transcribe English and other languages, and even detect language changes in an audio conversation. Among its advantages, we find:

The ability to interpret pauses in conversations
Using this understanding to add commas and periods appropriately based on the length of the pause

Whisper is a language model that serves as the basis for developing applications and resources. Businesses can connect their website to this template via its API to create transcription or translation tools.

Different versions of Whisper

There are different sizes of Whisper for different applications, ranging from a lightweight version with less than 1 GB of VRAM to a larger model with 1.55 trillion parameters and requirements of around 10 GB of VRAM.

How to use Whisper?

To use Whisper, you can go to its page on Github for advanced technical instructions, or go to the platform replicate.com/openai/whisper, which offers the use of Whisper and other AI models in a simple way . There you will be able to upload your audio files and select the model of your choice, including v3 in its different versions, although registration is required for more advanced use.

In short, Whisper is a major innovation in the field of audio transcription. With its use of artificial intelligence, it offers unparalleled accuracy and efficiency, making the task of transcription much easier and faster. Whether you’re a journalist, translator, or simply someone who needs to transcribe audio files regularly, Whisper is a tool worth trying.