Purfview
Purfview
I think `cpu_threads` doesn't do anything if on cuda and ` compute_type="bfloat16"` should be fastest for your GPUs. Dunno much about your multithreading issue.
> Do you have a sense of why the transcriptions are not really using Tensor Cores but only CUDA cores? With "bfloat16"?
On Windows "Hardware-accelerated GPU scheduling" is hyper option, dunno if there is such thing on Linux.
Is that model working with vanilla Whisper and word_timestamps=True?
Read there https://github.com/SYSTRAN/faster-whisper/issues/699
This repo doesn't have CLI, for CLI functionality look at these repos: https://github.com/Softcatala/whisper-ctranslate2 https://github.com/Purfview/whisper-standalone-win
>word_timestamps Supported. >highlight_words Supported. >max_words_per_line Not supported. EDIT: That's about https://github.com/Purfview/whisper-standalone-win
Can you share an audio sample to reproduce the issue?
> I am also seeing this issue. > ``` > [0.00s -> 29.20s] Turn on kitchen sink. Turn on kitchen sink. > ``` > > I get the same output...
> Do you have to install the cuBLAS/cuDNN libraries for CUDA 11, or will it also work with the CUDA 12 versions? You've to install the cuBLAS/cuDNN libraries for CUDA...