System Runtime

[GorinIlya]

System.Runtime.InteropServices.SEHException (0x80004005): External component has thrown an exception.
   at Whisper.net.Internals.ModelLoader.WhisperProcessorModelFileLoader.LoadNativeContext(INativeWhisper nativeWhisper)
   at Whisper.net.WhisperFactory..ctor(IWhisperProcessorModelLoader loader, Boolean delayInit)
   at Whisper.net.WhisperFactory.FromPath(String path, WhisperFactoryOptions options)
   at Whisper.net.WhisperFactory.FromPath(String path)
   at FlyleafLib.MediaPlayer.WhisperExecuter.Do(MemoryStream waveStream, TimeSpan chunkStart, TimeSpan chunkEnd, Int32 chunkCnt, CancellationToken token)+MoveNext() in C:\src\LLPlayer\FlyleafLib\MediaPlayer\SubtitlesASR.cs:line 312
   at System.Threading.Tasks.ValueTask`1.GetTaskForValueTaskSource(IValueTaskSource`1 t)
--- End of stack trace from previous location ---
   at FlyleafLib.MediaPlayer.AudioReader.<ReadAll>g__ProcessChunk|11_0(MemoryStream stream, Int64 endPts, <>c__DisplayClass11_0&) in C:\src\LLPlayer\FlyleafLib\MediaPlayer\SubtitlesASR.cs:line 593
   at FlyleafLib.MediaPlayer.AudioReader.ReadAll(TimeSpan curTime, Action`1 addSub, CancellationToken token) in C:\src\LLPlayer\FlyleafLib\MediaPlayer\SubtitlesASR.cs:line 550
   at FlyleafLib.MediaPlayer.SubtitlesASR.Open(Int32 subIndex, String url, Int32 streamIndex, TimeSpan curTime) in C:\src\LLPlayer\FlyleafLib\MediaPlayer\SubtitlesASR.cs:line 153
   at FlyleafLib.MediaPlayer.Subtitle.<>c__DisplayClass39_0.<EnableASR>b__0() in C:\src\LLPlayer\FlyleafLib\MediaPlayer\Subtitles.cs:line 649
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)
--- End of stack trace from previous location ---
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state)
   at System.Threading.Tasks.Task.ExecuteWithThreadLocal(Task& currentTaskSlot, Thread threadPoolThread) 

What`s wrong and how to fix it ?

[umlx5h]

@GorinIlya
Hello, Could you please complete the following template from issue bug report?

The version can be obtained from the `About` section of the settings.

- OS: [e.g. Windows 10, 11]
- CPU Architecture: [e.g. x64, x32, ARM64]
- LLPlayer version: [e.g. v0.0.1]
- LLPlayer commit version: [e.g. 86f063df2c56a198b204664951a6147ee739f1b7]

If the bug relates to `Whisper` / `ASR`, please complete the following.

- GPU: [e.g. NVIDIA RTX 4060]
- Whisper Hardware Options: [e.g. Cuda -> Cpu -> CpuNoAvx]
- [Microsoft Visual C++ Redistributable Version >=2022](https://learn.microsoft.com/en-us/cpp/windows/latest-supported-vc-redist?view=msvc-170#latest-microsoft-visual-c-redistributable-version) Installed: [e.g. True]

If you have time, it would be helpful if you could download the following, run the exe and post the results as well.
First you will be asked if you want to use CUDA, so please press n key.

https://github.com/umlx5h/LLPlayer/releases/download/v0.0.1/WhisperTesting.7z

[GorinIlya]

WITHOUT CUDA

Use CUDA? (y/N): N
[Error] whisper_init_from_file_with_params_no_state: loading model from 'ggml-tiny.bin'

[Error] whisper_init_with_params_no_state: use gpu    = 1

[Error] whisper_init_with_params_no_state: flash attn = 0

[Error] whisper_init_with_params_no_state: gpu_device = 0

[Error] whisper_init_with_params_no_state: dtw        = 0

[Info] ggml_vulkan: Found 1 Vulkan devices:

[Info] ggml_vulkan: 0 = NVIDIA GeForce RTX 4060 (NVIDIA) | uma: 0 | fp16: 1 | warp size: 32 | matrix cores: KHR_coopmat

[Error] whisper_init_with_params_no_state: devices    = 2

[Error] whisper_init_with_params_no_state: backends   = 2

[Error] whisper_model_load: loading model

[Error] whisper_model_load: n_vocab       = 51865

[Error] whisper_model_load: n_audio_ctx   = 1500

[Error] whisper_model_load: n_audio_state = 384

[Error] whisper_model_load: n_audio_head  = 6

[Error] whisper_model_load: n_audio_layer = 4

[Error] whisper_model_load: n_text_ctx    = 448

[Error] whisper_model_load: n_text_state  = 384

[Error] whisper_model_load: n_text_head   = 6

[Error] whisper_model_load: n_text_layer  = 4

[Error] whisper_model_load: n_mels        = 80

[Error] whisper_model_load: ftype         = 1

[Error] whisper_model_load: qntvr         = 0

[Error] whisper_model_load: type          = 1 (tiny)

[Error] whisper_model_load: adding 1608 extra tokens

[Error] whisper_model_load: n_langs       = 99

[Error] whisper_default_buffer_type: using device Vulkan0 (NVIDIA GeForce RTX 4060)

ggml_vulkan: Compiling shaders................................Done!
[Error] whisper_model_load:  Vulkan0 total size =    77.11 MB

[Error] whisper_model_load: model size    =   77.11 MB

#### Using CPU ####
[Error] whisper_backend_init_gpu: using Vulkan0 backend

[Error] whisper_init_state: kv self size  =    3.15 MB

[Error] whisper_init_state: kv cross size =    9.44 MB

[Error] whisper_init_state: kv pad  size  =    2.36 MB

[Error] whisper_init_state: compute buffer (conv)   =   14.15 MB

[Error] whisper_init_state: compute buffer (encode) =   64.79 MB

[Error] whisper_init_state: compute buffer (cross)  =    3.88 MB

[Error] whisper_init_state: compute buffer (decode) =   96.81 MB

[Error] whisper_full_with_state: auto-detected language: en (p = 0.993854)

00:00:00->00:00:12:  I believe that this nation should commit itself to achieving the goal before this decade is out of landing a man on the moon and returning him safely to the earth.
00:00:12->00:00:22:  No single space project in this period will be more impressive to mankind or more important for the long range exploration of space.

Success
Press Enter to exit

WITH CUDA:

Use CUDA? (y/N): Y
[Error] whisper_init_from_file_with_params_no_state: loading model from 'ggml-tiny.bin'

[Error] whisper_init_with_params_no_state: use gpu    = 1

[Error] whisper_init_with_params_no_state: flash attn = 0

[Error] whisper_init_with_params_no_state: gpu_device = 0

[Error] whisper_init_with_params_no_state: dtw        = 0

[Info] ggml_vulkan: Found 1 Vulkan devices:

[Info] ggml_vulkan: 0 = NVIDIA GeForce RTX 4060 (NVIDIA) | uma: 0 | fp16: 1 | warp size: 32 | matrix cores: KHR_coopmat

[Error] whisper_init_with_params_no_state: devices    = 2

[Error] whisper_init_with_params_no_state: backends   = 2

[Error] whisper_model_load: loading model

[Error] whisper_model_load: n_vocab       = 51865

[Error] whisper_model_load: n_audio_ctx   = 1500

[Error] whisper_model_load: n_audio_state = 384

[Error] whisper_model_load: n_audio_head  = 6

[Error] whisper_model_load: n_audio_layer = 4

[Error] whisper_model_load: n_text_ctx    = 448

[Error] whisper_model_load: n_text_state  = 384

[Error] whisper_model_load: n_text_head   = 6

[Error] whisper_model_load: n_text_layer  = 4

[Error] whisper_model_load: n_mels        = 80

[Error] whisper_model_load: ftype         = 1

[Error] whisper_model_load: qntvr         = 0

[Error] whisper_model_load: type          = 1 (tiny)

[Error] whisper_model_load: adding 1608 extra tokens

[Error] whisper_model_load: n_langs       = 99

[Error] whisper_default_buffer_type: using device Vulkan0 (NVIDIA GeForce RTX 4060)

ggml_vulkan: Compiling shaders................................Done!
[Error] whisper_model_load:  Vulkan0 total size =    77.11 MB

[Error] whisper_model_load: model size    =   77.11 MB

#### Using CUDA ####
[Error] whisper_backend_init_gpu: using Vulkan0 backend

[Error] whisper_init_state: kv self size  =    3.15 MB

[Error] whisper_init_state: kv cross size =    9.44 MB

[Error] whisper_init_state: kv pad  size  =    2.36 MB

[Error] whisper_init_state: compute buffer (conv)   =   14.15 MB

[Error] whisper_init_state: compute buffer (encode) =   64.79 MB

[Error] whisper_init_state: compute buffer (cross)  =    3.88 MB

[Error] whisper_init_state: compute buffer (decode) =   96.81 MB

[Error] whisper_full_with_state: auto-detected language: en (p = 0.993854)

00:00:00->00:00:12:  I believe that this nation should commit itself to achieving the goal before this decade is out of landing a man on the moon and returning him safely to the earth.
00:00:12->00:00:22:  No single space project in this period will be more impressive to mankind or more important for the long range exploration of space.

Success
Press Enter to exit

[umlx5h]

There was a problem with the runtime configuration of the sample program, which has been updated.

If you don't mind, could you please run it again with pressing n key?
https://github.com/umlx5h/LLPlayer/releases/download/v0.0.1/WhisperTesting.7z

It may be that for some reason it cannot be run if it is CPU in your environment.
What exactly CPU model are you using?

Please try running it with Vulkan only, or try CPUNoAvx only to see if it works.

Alternatively, as you seem to be using an NVIDIA GPU, I recommend that you install the CUDA driver from below and run it with CUDA, which is the fastest option.

https://developer.nvidia.com/cuda-downloads