I keep getting the error that i reached maximum content size with Whisper API. However, I do have Plus version and have more capital available in the account to be able to make more calls. It’s been 3 days in a row I am receiving this error, so it is no longer “wait and your IP will be allowed” type of a thing. Can anybody help with this?
No idea at this point, but I am curious what is the size limit for audio file?
BTW, I was able to do successful calls in C# with previous endpoint/api, but not with the latest one where you indicate “whisper-1” model as parameter…
25 Meg limit on the file. Use a tool to break it into smaller pieces and feed in these smaller pieces and stick them together. See the docs for tools that can do this without breaking across a word.
My learnings so far are, that 25 Meg is about 22 Minutes length if you have an MP3-File with 32 Khz. I cut longer files with either Adobe Audition or the free Audacity.
By the way: whisper does a good job in understanding two languages at once. I had an interview with a german and an english speaker and content was really good.
I’m using the Node API and can only seem to send .mp3 files of 8 MB or shorter, even though all the docs say I should be able to send files up to 25 MB in length.
Anyone else seen this in node?
I’m sending the file via:
const transcription = await AI.createTranscription(fs.createReadStream("foo.mp3"), "whisper-1");
You need to set the Axios config object in Node:
const resp = await openai.createTranscription(
fs.createReadStream("file.ogg"),
"whisper-1",
undefined,
undefined,
undefined,
undefined,
{
maxBodyLength: 25 * 1024 * 1024,
},
)
Thanks, it saved the day. Why openai Whisper doc doesn’t mention about maxBodyLength
? Curious where did you find it.
check this
Saved the day, buddy! TYSM
OpenAI folks, this should be in the docs, or better yet not even be needed at all