Bill Simmons, founder of the Spotify-owned podcast network The Ringer, hinted in May that Spotify also planned to use AI trained on the voices of hosts to generate targeted ads for users.
This is what is really messed up to me (emphasis mine).
this post was submitted on 27 Sep 2023
73 points (83.5% liked)
Technology
59461 readers
4545 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
IANAL but pretty sure that's illegal.
It violates the right of publicity (which is either state-specific or common law) which is explicitly recognized to include commercial control of the likeness of oneself. That most certainly includes someone's voice. There's probably an argument that for AI generated translations some podcast producers have an interest in that they might wanna agree (broader reach), but with ads that is pretty cut and dry. At that point, I kinda hope the legal department knows this, because otherwise this could do incredible damage to their PR (as anything illegal does).
Stop posting business insider, almost every legitimate thing there you can find covered better by other sources and all exclusives are probably improperly sourced
It sounded good so I tried this and it's so so at best.
First off, I pay for no ads. To me, a fake DJ attempting to convince me it's a black man talking nonsense is just as bad or worse than ads.
Second, the music choices are early days Pandora bad. It starts off okay then a few songs later it's off on some tangent with no rhyme or reason I can understand.
Eerie? This is awesome!
It was until you hit the part in the article about using AI voice generation to target ads. When election season rolls around, we'll get all sorts of clips of people endorsing Trump without their knowledge.
If 2016 was the post-truth election, 2024 is going be the post-reality era.
Honestly, 2024 seems more like an attempt to return to sanity while the elites are trying to learn how to use their new toys. 2026 midterms will be the beta test and 2028 is going to be the real shit show.
That could already happen with AI technology from months ago. This is not news.
The targeted ads will be that if you follow specific broadcasts, you will get an higher chance to get an ad voiced by a person in that podcast. AI has little to do here, could have always been done.
Unfortunately its use in the US politics its inevitable, with what I am understanding since i've started to follow it.
The dangerous part is that it sounds like the ad is generated by Spotify.
In the "tech already exists" scenario, an advertising firm will have to guess at the voice people want to hear and submit their own audio files to advertise.
If this works the way it sounds, the ad firm sends in some text and Spotify generates a voice ad based on who you listen to. Less effort on the advertising firm and far more targeted.
Both are bad. The second is worse!
And frankly, it's another reason not to support Spotify. You are supposed to be paying to listen to music, not support research into this bullshit advertising.
There's a difference between being able to technically do this, and deploying it as a product with legal checks and scaled up infrastructure.
Since the article was about generative voice models, I assumed the ad part was also about that, but you're right it might just be using voice recognition.
Yeah, this sounds like a really GOOD use of AI.
The part that the news line wants to point at probably is where it might be planned to use to create ads using podcaster's voices. I don't see what's the issue there is, now the podcasters can be paid to get their voice imitated for an ad they don't even have to practice the lines for.
But do they have control over what's in the ads? I know I wouldn't want my voice used to promote things without my knowledge
I am very sure that any company that doesn't want to risk legal battles would respect your choice. If the usage of your voice is behind a contract, you can definitely ask what it will be about and what will be said with the freedom to refuse it.
Its the same situation of using without AI another company's brand to sponsor something else.
And will they really be paid or provided a choice?
I fully agree that this is a great use for ai but voices are a part of peoples identity we should be careful before selling it is normalized like taxes
I think so. Unless they eventually want to end up in a legal battle.
This is the best summary I could come up with:
Spotify is rolling out a new AI feature mimicking the original voice of podcasters and translating them into additional languages for a more "authentic listening experience," the company announced Monday in a blog post.
"A podcast episode originally recorded in English can now be available in other languages while keeping the speaker's distinctive speech characteristics," per Spotify's blog.
Spotify is working with podcasters like Dax Shepard, Monica Padman, Lex Fridman, Bill Simmons and Steven Bartlett to produce voice translations for specific episodes in languages such as Spanish, French, and German.
Bill Simmons, founder of the Spotify-owned podcast network The Ringer, hinted in May that Spotify also planned to use AI trained on the voices of hosts to generate targeted ads for users.
There have been mounting concerns in the entertainment industry about the impact of AI being used to clone the voices of musicians, actors, and presenters without their consent or providing financial compensation.
Actor and comedian Stephen Fry also recently said that an AI had ingested his readings of the "Harry Potter" books and subsequently recreated his voice.
The original article contains 285 words, the summary contains 178 words. Saved 38%. I'm a bot and I'm open source!