thewh1teagle / vibe

Transcribe on your own!
https://thewh1teagle.github.io/vibe/
MIT License
907 stars 55 forks source link

Bug: Trying to transcrbe, second chance #155

Closed nonZero closed 1 month ago

nonZero commented 3 months ago

What happened?

an error dialog

Steps to reproduce

selected a file, clicked on transcribe

What OS are you seeing the problem on?

Linux

Relevant log output

failed to open model

Caused by:
    Failed to create a new whisper context.

Location:
    core/src/model.rs:33:6
App Version: 2.0.6
Commit Hash: 23815ae6f983e692631d1eac72ccfefa06b90199
Arch: x86_64
Platform: linux
Kernel Version: Rolling Release
OS: linux
OS Version: Rolling Release
Cuda Version: 12.4.1
Models: ggml-medium.bin
Default Model: ggml-medium.bin"

CPU feature detection is not supported on this architecture.
thewh1teagle commented 3 months ago

Hey! Thanks for reporting! It seems like the model file is invalid. Can you check the sha256 of the file ggml-medium.bin? You can open the models folder through the settings.

sha256sum <path>

In addition you can try re-download the model by resetting the app through the setting.

nonZero commented 3 months ago

6c14d5adee5f86394037b4e4e8b59f1673b6cee10e3cf0b11bbdbee79c156208 ggml-medium.bin

thewh1teagle commented 3 months ago

6c14d5adee5f86394037b4e4e8b59f1673b6cee10e3cf0b11bbdbee79c156208 ggml-medium.bin

Thanks! That looks valid.

I'm not sure what's the error, I tried latest version on ubuntu 24.04 and it worked. Can you try to start it from the console with debug mode, then try to transcribe, and paste the log here?

In terminal:

export RUST_LOG="whisper_rs=debug,vibe=debug"
vibe

You can also run it entirely from the terminal

export RUST_LOG="whisper_rs=debug,vibe=debug"
vibe --model "ggml-medium.bin" --file path.wav
thewh1teagle commented 1 month ago

Please try the latest version, should be fixed.