Maximum length of transcription

[OmerIed]

Does whisper has a maximum length for an output (for a single inference of an input of 30 seconds)?
If there is a maximum, what is it? is it defined by number of tokens/characters?

[ryanheise]

model.dims.n_text_ctx indicates 448 tokens for a 30 second window, but half of that is reserved by the prompt, so that leaves 224 tokens for output.