whisper.cpp config model dropdown UI when manually adding models

[Firemission1343]

I've found a ggml model for distil-large-v3.5 on huggingface that I wanted to test with the whisper.cpp engine and found a UI issue for the whisiper.cpp config model dropdown will not show any manually added custom named models.

My current workaround is renaming a model that already exist in the directory, example; distil-large-v3.5 renamed into ggml-largev3.bin, and that seems to work.

Would be useful to have if later on for future support of other ggml models like parakeet-tdt-0.6b-v2

Also I wanted to add that this project is extremely helpful for me and awesome!

[olbldwaste]

I ran into the same issue. LLPlayer relies on the Whisper.net library, but unfortunately it only comes with a fixed set of speech recognition models baked in.

Here’s the list of models Whisper.net currently supports:

namespace Whisper.net.Ggml;
public enum GgmlType
{
    Tiny,
    TinyEn,
    Base,
    BaseEn,
    Small,
    SmallEn,
    Medium,
    MediumEn,
    LargeV1,
    LargeV2,
    LargeV3,
    LargeV3Turbo
}

Hopefully the developers figure out a way to avoid this, because LLPlayer itself can actually work with any model. Right now, the way Whisper.net handles it feels like a weird half‑measure, more of a lazy boney than a proper implementation.