ElevenLabs just got nuked by open source

Links: Qwen3-TTS demo app: That time my voice was cloned without consent:

I agree that eleven labs is getting nuked by opensource, but not for this reason. Voice cloning is a very small piece of ElevenLabs’ platform. Projects like and livekit are much more likely to cause issues for them long term, imho.

You say obviously that doesn't sound like you're voice but it sounds exactly like your voice

have you tried real-time speech-to-speech?

elevenlabs getting nuked by open source? more like they’re finally exposed. open source always wins in the end.

oh boy... stuff like this is going to make people's live so much better... ... no... wait... I misspelled worse there.

You just randomly popped up in my feed and it freaked me out at first because I thought it was a deepfake video of me. Hello, doppelgänger! Oddly enough, I'm also from STL.

#stl represent!

There are many great open source TTS at this point. ElevenLabs will serve enterprise - making money from those who have money and want professional setup, security, support; not from indie hackers who will run it on a raspberry pi.

I spent the entire video waiting for the "and if you're not convinced... you've been listening to it this whole time". Extremely disappointed.

I've got such a shit and hated voice, no one would want to clone it 😁

"Default distrust." That's rapidly becoming the new requirement for modern media literacy. In other words, don't trust anything you see, unless it's cryptographically signed by the creator. Technology has outpaced this infrastructure, so until this happens, we're in the wild west.

chinese models will soon cross western models capabilities

If you haven't seen it yet, @kyutai_labs just released models that do this too and it's also entirely FOSS (MIT licensed). Their demo site is at . You can record your own voice and have it speak back to you.

A major problem with this are phone scammers are going to trick family members more easily

I was waiting for the plot twist on this video: “And yeah, the video you just watched was entirely generated by AI as well.”

While nuking a business usually isn't a viable long-term strategy, I'm curious to see how ElevenLabs chooses to respond to this setback.

I just got it working via the example code and assets. It sounds exactly like the input audio. Tone and all. The world has changed, I think we all need to accept that and begin to change our understanding of data and human interactions. Audio is in the video 👇🏽

"nuked by open source" the elevenlabs pricing page just felt a disturbance in the force. this is how every ai moat dies. slowly, then all at once.

get hair like jeff's

Thank you algo for putting this on my For You page. Qwen have been great with their open source models

The anons won in the end! Everyone will be everyone! Everyone will be anon! Identity is meaningless! Only what you say matters not who is saying it! Every post will be judged on its own merits not the namefag/voicefag/facefag identity!

I get your gripe about your “VOICE” that’s why I suggested to Sam Altman. That we should have a database, we can monetize our own voice and apply to copyright system to our Voice and people can choose if their AI companion will use our voice, but allow us to receive royalties.

Banks have voice auth now. Considering how crappy voice is over the phone, this could make their authentication worthless.

Jeff didn’t want us to know this entire video was AI

nice Sneakers reference

Identify theft and impersonation is a serious issue, but as someone whose voice gives out easily being able to use this to replicate my own voice to create audio would be a huge relief.

This video is all about open source voice cloning which has been passable for about a year. Not about ElevenLabs. Want my 4 minutes back.

They’re doing 300m ARR , I think they’ll be fine!

ElevenLabs have well and truly moved on from this basic level (I am a daily driver) but yes its an interesting development I have been looking into for quick edge/field use. Even elevenLabs has yet to cross the threshold into truly pro-podcast aka "radiolab" territory, though its good for video dubbing, promos and little corrections in post (just completed an extensive A/B on this)

Good. I tried to use Elevenlabs couple times by they kept refusing to process anything from me because I connected via VPN lol.

I think they need to be more worried about Nividias open source voice model 😅

cool stuff

I suspect Voight-Kampff Tests will be performed a lot more and very soon.

ElevenLabs' pricing can't be justified. I can achieve reasonable quality locally with similar speeds on 8-year old hardware for pennies.

@grok give me transcription for this video