RuntimeError: CUDA failed with error CUDA driver version is insufficient for CUDA runtime version

[olawalejuwonm]

Steps to reproduce

I followed the guide here https://github.com/HenestrosaDev/audiotext#set-up-the-project-locally

With these options:

Expected behaviour

It should transcribe the audio file selected

Actual behaviour

Errors out: RuntimeError: CUDA failed with error CUDA driver version is insufficient for CUDA runtime version

(If the library threw an exception, paste the full stack trace here)

Traceback (most recent call last):
File "C:\Users\HP\Documents\GitHub\Tarm\audiotext\src\handlers\whisperx_handler.py", line 30, in transcribe_file
model = whisperx.load_model(
^^^^^^^^^^^^^^^^^^^^
File "C:\Users\HP\Documents\GitHub\Tarm\audiotext.venv\Lib\site-packages\whisperx\asr.py", line 288, in load_model
model = model or WhisperModel(whisper_arch,
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\HP\Documents\GitHub\Tarm\audiotext.venv\Lib\site-packages\faster_whisper\transcribe.py", line 133, in init
self.model = ctranslate2.models.Whisper(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: CUDA failed with error CUDA driver version is insufficient for CUDA runtime version

System information

(Delete all statements that don't apply.)

• System: "Windows 10 x64"
• System language: (Please, indicate the region as well): English
• Audiotext version: v2.2.3

[HenestrosaDev]

Can you share your CPU and GPU models? I don't understand why you get the CUDA error having the Use CPU option checked. It would only make sense if you were using the GPU to generate the transcription.

[olawalejuwonm]

How do I share that?

…

On Tue, 16 Jul 2024, 22:13 HenestrosaDev, ***@***.***> wrote: Can you share your CPU and GPU models? I don't understand why you get the CUDA error having the Use CPU option checked. It would only make sense if you were using the GPU to generate the transcription. — Reply to this email directly, view it on GitHub <#40 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMSQ2BGZGRDLYGH2HHBUHHTZMWEH3AVCNFSM6AAAAABK6MZO72VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMZRHAZTQMZWG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[HenestrosaDev]

Follow these steps on Windows 10 to check your CPU model:

1. Press Windows + I and select "System".
2. Scroll down the sidebar to the bottom and click "About".
3. Copy and paste here the "Processor".

To find your GPU model on Windows 10, do the following:

1. In the search box on the taskbar, type "System information" and open it.
2. In the system Information window, expand the "Components" option.
3. Click on "Display".
4. Copy and paste the value of "Name".

[olawalejuwonm]

Follow these steps on Windows 10 to check your CPU model:

1. Press Windows + I and select "System".
2. Scroll down the sidebar to the bottom and click "About".
3. Copy and paste here the "Processor".

To find your GPU model on Windows 10, do the following:

1. In the search box on the taskbar, type "System information" and open it.
2. In the system Information window, expand the "Components" option.
3. Click on "Display".
4. Copy and paste the value of "Name".

CPU Model: Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz 2.40 GHz
GPU Model: Intel(R) HD Graphics 520

[HenestrosaDev]

1. Can you change the compute type setting to int8?
2. If the problem still persists after doing 1), please share the contents of your config.ini file. It's located in the root of the audiotext-v2.2.3 folder.

[olawalejuwonm]

1. Can you change the compute type setting to int8?
2. If the problem still persists after doing 1), please share the contents of your config.ini file. It's located in the root of the audiotext-v2.2.3 folder.

Option 1 doesn't work.
2. [whisperx]
model_size = tiny
batch_size = 8
compute_type = float16
use_cpu = False
can_use_gpu = False
output_file_types = txt

[google_api]
api_key =

[subtitles]
highlight_words = False
max_line_width = 42
max_line_count = 2

[system]
appearance_mode = System

[HenestrosaDev]

Change use_cpu = False to use_cpu = True. This should fix the problem. I have to check why the value doesn't change to True on start.

[olawalejuwonm]

Yes, it works!

…

On Fri, Jul 19, 2024 at 9:39 PM HenestrosaDev ***@***.***> wrote: Change use_cpu = False to use_cpu = True. This should fix the problem. I have to check why the value doesn't change to True on start. — Reply to this email directly, view it on GitHub <#40 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMSQ2BDLWPERXOATZTG53D3ZNF2OZAVCNFSM6AAAAABK6MZO72VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBQGA3TCMJXGA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[olawalejuwonm]

But I have another issue, I will open in another issue