Faster-Whisper GPU Setup on Windows 11 | CUDA 12, cuDNN 9, cuBLAS Fix

Published 14/07/2026

Description

Install Faster-Whisper with GPU acceleration on Windows 11 using CUDA 12 and cuDNN 9. In this tutorial, I show how to set up Faster-Whisper with Python, explain why GPU mode fails with cublas64_12.dll, and demonstrate how to get NVIDIA CUDA working for fast local speech-to-text transcription. Contents: 0:00 - Introduction 0:10 - Description 1:01 - GitHub Base 1:25 - Python PIP Install 2:02 - Model Retrieval 5:07 - Simple Run GPU Style 7:06 - Requirements 7:39 - CUDA 12.x 9:04 - cuBLAS for CUDA 12.x 10:17 - cuDNN 9 for CUDA 12.x 11:06 - Successfull Run GPU Style 11:28 - Model Performance 13:33 - Live Transcribe 14:13 - Conclusion You’ll learn how to: Install Faster-Whisper with pip. Download Faster-Whisper models from Hugging Face. Run Faster-Whisper on CPU and GPU. Fix the cublas64_12.dll error on Windows 11. Understand why CUDA 12 already includes the required cuBLAS 12 library. Install cuDNN 9 for CUDA 12. Verify GPU usage and compare CPU vs GPU transcription speed. See why distilled models can be dramatically faster on GPU. If you are trying to run local AI transcription with an NVIDIA GPU and Faster-Whisper keeps falling back to CPU, this walkthrough shows the full setup and the exact issue behind the missing cuBLAS error. If you're interested in AI, Python, speech recognition, Whisper, Faster-Whisper, CTranslate2, and real-time audio processing, this video will help you get started quickly. Don’t forget to like, subscribe, and leave a comment if you want more videos about AI tooling, Python automation, and practical tech projects. #FasterWhisper #CUDA12 #SpeechToText GitHub : https://l.samayas.eu/llkVBBno7V Follow me on X: https://l.samayas.eu/jo3eaYEf3E?P=YT&V=136 Buy me a Coffee : https://l.samayas.eu/jX6xL7EWhd?P=YT&V=136 Studio Equipment Camera Sony Alpha ZV-E10 II: https://l.samayas.eu/k7y95TREl0?P=YT&V=136 Secondary Camera Obsbot Tail Air: https://l.samayas.eu/iFQu9WfmCw?P=YT&V=136 Elgato Prompter: