Transcription is definitely the worst part of doing an interview. I love talking to people who have something to say, but then transcribing it is a very slow process. You should listen to the audio again and type quickly. That’s the manual process. Fortunately, there are programs that convert audio to text. What’s happening? Basically they either get paid or they work very badly.

But artificial intelligence came to fix things. And we’re talking specifically about Whisper from the creators of ChatGPT.

It may not be as well known as a chat tool or a rendering tool, but Whisper is an audio-to-text converter that works great. And in all languages. Because here is exactly what I like the most; it also offers a high level of accuracy in Spanish.

It’s true that we have more alternatives in English but Most of the free tools in Spanish have failed badly. Things change with Whisper.

Whisper is described as an “automatic speech recognition (ASR) system.” A model trained with over 680,000 hours of data. And since it’s open, we can access it on the web.

In Replicate we have the option to access the large v2 model. No need to install anything, we just upload the audio file We want and return text that changes depending on the length and size of the audio file after a while.

The result is very accurate, although the audio is a complex call. Additionally, Whisper allows you to choose the result format in plain text or subtitles.

it is not infallible. Whisper is wrong in technical words such as proper names and is not reserved by the addressees for now.. But it works very quickly and puts pauses in speech where they belong.

It is true that I should then do a good review at the fiction level; Check if the sentences are touching sentences and if no sections are skipped. Not currently uploading and streaming audio, but Whisper is an incredible leap forward. saved me hours of work.

So far I have been using Otter.ai for English text, but from now on Whisper is the default voice-to-text translation engine that I would definitely recommend. With all eyes on ChatGPT, I’m looking forward to the new version of Whisper.

Image | Bogomil Mikhailov

on Xataka | Convert Audio to Text: 17 Free Tools