We have seen AI being used more and more for creating human-sounding voices, including their use in creating “fake news”. Today, the audio streaming service Spotify announced it is launching an AI-generated voice feature, but one that is supposed to be used for good, not evil.
In a blog post, Spotify revealed it is beginning to pilot its Voice Translation AI feature in a number of its podcasts. The feature will let people in other countries and languages listen to native versions of those podcasts, with the host speaking as though they were using those languages, instead of different people speaking the words from the original host.
Spotify stated it is using technology from OpenAI, which just announced it is adding voice chat options to its ChatGPT mobile apps. Spotify stated:
This Spotify-developed tool leverages the latest innovations—one of which is OpenAI’s newly released voice generation technology—to match the original speaker’s style, making for a more authentic listening experience that sounds more personal and natural than traditional dubbing. A podcast episode originally recorded in English can now be available in other languages while keeping the speaker’s distinctive speech characteristics.
The new technology is slowly being rolled out, with Spotify stating that well-known podcasters like Dax Shepard, Monica Padman, Lex Fridman, Bill Simmons, and Steven Bartlett have worked with the streamer to use their voices for this program. It will be used to create Spanish, French, and German versions of their voices for select podcast episodes, along with future episodes of their podcasts.
Spotify listeners can access these translated podcasts on the Now Playing View right now. You can also head over to the dedicated Voice Translations Hub to listen to these episodes. More will be added to the hub in the future, including episodes of the upcoming and unnamed podcast from former Daily Show host Trevor Noah.