HATEIt is written in capital letters, bold, and links to the definition of “hate” in RAE’s WhatsApp audios. I understand they are used, they are very useful when you want to say something “”””quickly”””””””””””””””””””””””” “”””””””””””””””””””””””””” “”””””””””””””””” and you have your hands full, but if we consider the recipient, the person who has to listen to your podcast, the truth is that audio presents some problems that cannot be ignored: to know all the content you have to listen, it cannot be searched, it is not short and above all it is not always possible to listen to a voice.

In some cases we have had the sound yes, sound no debate and today I was blessed with the arrival of the last one, as if an all-powerful being called God, called the engineer from Meta, heard my prayers. Great WhatsApp feature. I’m talking, of course, about automatic transcription, a function that I’ve already been able to experience firsthand. And no, I won’t listen to audio again.

sound problem. There are several, but the most important one is that voice messages greatly benefit the sender, not the receiver. A short voice message may well be a text message of four or five words (typed in two seconds). A long voice message can be a call that allows for instant two-way communication. The problem with that five-minute voice can definitely be learned in a 30-second conversation.

Also, unlike text, voices allow for unnecessary digressions resulting from improvisation and unstructured speech. This causes the main message to be diluted with main lines and digressions. eeeeeh, Hmma sneeze that is very pleasant to hear, the sound of a car or the wind, etc. Not to mention that they cannot be searched using the WhatsApp search engine.

It also has its advantages, as we discuss in this article. If you want, you can listen below. Only four minutes, very comfortable, just like WhatsApp sounds.

Transcription. To at least partially solve this problem, WhatsApp introduced automatic transcription. It’s not publicly available and is being rolled out gradually, so it may not be available on your device yet. Also, if it is, it won’t be enabled by default. To check, go to Settings > Chats and see if an option called “Voicemail Transcription” appears. If it appears enable it and download the language pack, it is around 130 MB. Otherwise we will have to wait.

Enabling WhatsApp voice transcription | Image: Xataka

How does it work? After configuring and enabling transcription, we should expect to receive audio via WhatsApp. When doing this, simply press and hold the message, then click on the three dots and finally click on “Transcribe”. WhatsApp will start converting the audio to text and display it below like in a small mini section.

It would be a success if the transcription were automatic and/or had fewer steps, but as a first approximation it’s great. It takes approximately four seconds to transcribe 30 seconds of audio. It lasted about 10-15 seconds, for a longer period of about four minutes.

Transcription is perfect by pushing for a more neutral accent and speaking slower | Image: Xataka

How is the performance? It’s not perfect by any means, but it does it well. I have a pretty strong Cordoban accent, I speak fast (you can watch any Xataka TikTok to make up your own mind), and WhatsApp can transcribe my voices without any problems. Sometimes, especially with some vowel vowels and quick connections between words, one of the words fails or is skipped, but this does not interfere with understanding the overall message.

I tried the system with some friends with different accents and the experience was surprisingly positive. It’s not perfect in any situation, but it’s more than accurate in any case, and it’s enough to know the content of the audio without listening to it.

It eats up some words when quickly chained with the previous ones but it works | Image: Xataka

WhatsApp eats some words when pronounced quickly and as usual trademarks, product names etc. When used, it largely fails. WhatsApp, which transcribed a conversation about 3D printers, could not understand words such as Prusa or BambuLab, but what was said is completely normal. What he understands without any problems is insult and swearing.

And indexed, half. Voices are not automatically transcribed, but when we transcribe them, that transcription remains in the chat and is treated as text by WhatsApp. What does this mean? We can use the search engine to find sounds. On one condition.

The transcript is not displayed in full, but only the first three lines are visible by default. In order for us to find the transcription using the search engine, the word we are looking for must appear in that preview. If we search for the word “center” we can find this message because this word appears in the preview, but if we search for the word “blue” which appears at the end we will not find it.

On the left is the indexed and searchable preview. At right, full transcript. Content visible after preview cannot be found in search engine | Image: Xataka

It’s not perfect but thank you. Far from being the spectacular transcription that tools like Whisper can achieve, what WhatsApp has achieved is more than enough to reduce audio friction. Listening to them will still be the best way to learn the entire content because of the intonation and emotions that only the voice can express, but transcription will allow us to do a vertical reading, preserve the general idea and record it if we wish. , listening to an unsolicited podcast.

Cover image | Xataka

in Xataka | My main WhatsApp group is now a “Community”. It was the best thing that ever happened to him