Ame on Nostr: So today I learned that there are open source, multilingual, fast, single-GPU models ...
So today I learned that there are open source, multilingual, fast, single-GPU models that will let you transcribe hours of audio in any language you want. Useful for work, but the quality of the results implies that you could set up a pipeline and give Vtubers in Japan immediate access to the EN market, with a few seconds of input lag, for basically free.
Published at
2024-07-27 08:07:21Event JSON
{
"id": "02cf20de66b1426cd612c0d9a46fd89ce113070e3a45f7a2f4a2fe4e18d106c0",
"pubkey": "7271f65788cf3d629727a54de8a74cfaa69f3ee1232c7dfc6bcb8059d28467e1",
"created_at": 1722067641,
"kind": 1,
"tags": [
[
"proxy",
"https://poa.st/objects/72a8dd81-0961-4de6-90c4-de19aa5181bc",
"activitypub"
]
],
"content": "So today I learned that there are open source, multilingual, fast, single-GPU models that will let you transcribe hours of audio in any language you want. Useful for work, but the quality of the results implies that you could set up a pipeline and give Vtubers in Japan immediate access to the EN market, with a few seconds of input lag, for basically free.",
"sig": "5d271ae6b794f078b3d508d76456aa4cd171c8752d2cac343e802fc7a4a364da439adf7f83a6ff3be3bd1d477d1d2f864de6a84f6e4882bd6420b78f9e7b5bf6"
}