正在加载视频...
视频加载失败
Over the weekend, Kled received a complex multichannel Korean audio task from one of the frontier labs. Within 48 hours, it was live in the Kled app. We are now sourcing thousands of Korean speakers to produce millions of hours of multichannel audio. The spec calls for per-speaker isolated... show more
84,242 次观看 • 4 天前 •via X (Twitter)
26 条评论

our team will simply out hustle any other provider in the market

Labs with a custom spec can explore the Kled marketplace and submit a request today. Our team reviews every submission and follows up directly with qualified labs:

we are Amazon Prime for frontier labs 👀

annyeonghaseyo

🔥🔥

upwards and onwards

Very cool

The 5-hour cap per speaker is interesting. Wonder why they picked that.

Persistent speaker IDs are interesting. What are they trying to teach the model?

a little dystopian ngl, but useful for real group chats

48 hours from lab request to phone task is pretty wild lol

Turning lab specs into something anyone can record on their phone is pretty clever.

🇰🇷🇰🇷🇰🇷

the first batch might be more interesting than the launch itself. wonder what comes next

Sick! Hope you recruit Korean Elvis

Quick math: with a five-hour cap per speaker, one million hours would need about 200,000 speakers. Is "millions of hours" counting each channel separately, or are you recruiting far beyond "thousands"?

Team so cracked. Patel mode

Patel mode

Persistent speaker IDs across recordings plus isolated per-speaker channels also make a very clean voice dataset. Does the documented consent cover voice cloning, or only speech recognition?

theres something so funny about a frontier lab handing kled a spec that specific and kled just having it live in two days like the rest of us are still waiting on a software update

Glad they included consent for training. Should be a given tbh.

This is cool lol

🇰🇷🇰🇷🇰🇷🇰🇷🇰🇷

custom data collection this fast could get researchers trying some pretty weird ideas lol

How does this help models?

getting AI to handle people talking over each other is harder than it sounds

