aaron on Nostr: So I’ve just transcribed something using #OpenAI’s #whisper model and, to my ...
So I’ve just transcribed something using #OpenAI’s #whisper model and, to my surprise, it added the string “Untertitelung des ZDF für funk, 2017 Untertitel von Stephanie Geiges” at the end. This is an interesting hint at the training material that was used. Turns out this issue isn’t too uncommon:
https://github.com/openai/whisper/discussions/928 
Published at
2024-09-12 19:15:47Event JSON
{
"id": "ac3239efb0ab21bdff32bd271b380ecea63c8ed23299801854c08ccf29e66de3",
"pubkey": "79bc9bbb03b46296fa8d52d6e7b6d017e4d586956cbc5455063f96ff35c38480",
"created_at": 1726168547,
"kind": 1,
"tags": [
[
"t",
"openai"
],
[
"t",
"whisper"
],
[
"proxy",
"https://mastodon.social/users/aaronk6/statuses/113126181927278313",
"activitypub"
]
],
"content": "So I’ve just transcribed something using #OpenAI’s #whisper model and, to my surprise, it added the string “Untertitelung des ZDF für funk, 2017 Untertitel von Stephanie Geiges” at the end. This is an interesting hint at the training material that was used. Turns out this issue isn’t too uncommon: https://github.com/openai/whisper/discussions/928\n\nhttps://files.mastodon.social/media_attachments/files/113/126/181/649/048/664/original/bfe701e9bed7f4a2.png",
"sig": "cdd517b6cf7d3d2dfac1d8c09c22082ac7052114ac95d195d3960a012eeba2a3bc6b80c26c8f44452ad3fbdfc55c0e1de1fc64293dbdc737093a03fa7fedff4a"
}
