Closed FredrikKarlssonSpeech closed 3 months ago
Hm... Running commands works fine. Can you try running same commands as below?
julia> using Whisper, Flux
julia> values(Whisper.LANGUAGES)
ValueIterator for a OrderedCollections.OrderedDict{String, String} with 99 entries. Values:
"english"
"chinese"
"german"
"spanish"
"russian"
"korean"
"french"
"japanese"
"portuguese"
...
julia> Whisper.transcribe(
"./samples/jfk.flac", "./output.srt";
model_name="tiny", language="english", dev=cpu, precision=f32)
[ Info: Running on `cpu` device at `f32` precision.
[ Info: Cache directory for weights: /home/pxl-th/.cache/Whisper.jl.
[ Info: Downloading weights: https://openaipublic.azureedge.net/main/whisper/models/65147644a518d12f04e32d6f3b26facc3f8dd46e5390956a9424a650c0ce22b9/tiny.pt
[ Info: Downloading tokens file: /home/pxl-th/.cache/Whisper.jl/multilingual.tiktoken
[ Info: Saving results to: `./output.srt`.
Sorry, this was an issue raised by me loading an identically named package instead of this repo code. Sorry.
I just tried to get this package to work, but it fails on the first attempt
Also, finding out available models do not work: