May 1, 2025
Trending News

Spotify uses Whisper to dub podcasts

  • September 25, 2023
  • 0

Spotify just made a master move. Sure, they were inspired by other more or less related actions we’ve seen in the past, such as plans to automatically translate

Spotify uses Whisper to dub podcasts

Spotify just made a master move. Sure, they were inspired by other more or less related actions we’ve seen in the past, such as plans to automatically translate YouTube videos and also Apple’s audiobooks, but that doesn’t detract from its strengths at all. On the contrary, it seems to me that this is the smartest move by the company in a long time, and if it does not get a quick response from its main competitors, it could be the difference when the user does not know whether to decide on one platform or the other.

Podcasts, as we’ve mentioned on occasion, live a second youth which is actually even more explosive than the first one when Apple started popularizing the format with the iPod. So, after trying to compete (with very little success) based on the exclusivity of their catalogs (which movie and series services do), streaming music services decided to take advantage of this resurgence and compete with their proposition. .

In this context, Spotify has been investing in this type of content for a long time, though it’s true that it announced cuts to its podcast division a year ago, and that we’ve previously verified that some grandiose announcements of exclusive content didn’t materialize. However, it is undeniable that today its offer is quite interesting in this sense and that it is already more than enough reason to choose Spotify over the competition.

Spotify uses Whisper to dub podcasts

Now, with the announcement made by the company through its blog, we find a winning movement in this sense, namely Spotify uses OpenAI Whisper along with proprietary technology to transcribe podcasts into other languages. In case you don’t know, Whisper is an AI based voice recognition model. The most common use is audio transcription, and once the text is already transcribed, it is translated and then a text to speech solution, trained with the original voice, is used to replicate the original content, but in the chosen output language. replicating the voice of the creator.

In the first tests of this new feature, Spotify worked with several creators, so the first dubbed content using this system was already available at the same time as the announcement of this new feature. They are the following:

  1. The Lex Fridman Podcast – “Interview with Yuval Noah Harari”
  2. The armchair expert : “Kristen Bell Returns by the Grace of God”
  3. CEO Diary with Steven Bartlett – “Interview with Dr. Mindy Pelz”

The originals were recorded in English, but thanks to this new Spotify proposal They are now available in Spanishthe French and German versions will be added in the coming days.

The result isn’t perfect, but it’s much more natural than other text-to-speech proposals and provides a virtually global reach for podcasts uploaded to the platform because it removes the language barrier to accessing podcasts. So the key now will be the pace at which more translated versions from more languages ​​are published. If Spotify acts quickly on this, it will score a big goal in its favor.

Source: Muy Computer

Leave a Reply

Your email address will not be published. Required fields are marked *