Const-me / Whisper

High-performance GPGPU inference of OpenAI's Whisper automatic speech recognition (ASR) model
Mozilla Public License 2.0
8.41k stars 721 forks source link

Unable to Batch Process Tasks #81

Open canaskpw opened 1 year ago

canaskpw commented 1 year ago

Title: Unable to Batch Process Tasks

Content: Hello Mr. Konstantin,

It's great to come across your Whisper project, which allows us to efficiently transcribe videos into text. However, currently, WhisperDesktop.exe (version 1.8.0.0) is unable to batch process tasks. After completing one task, we have to manually select the next one. I don't have experience in developing graphical software on Windows systems, but based on my limited scripting experience, if we want to batch process tasks serially, we can save the absolute paths of the selected files into a variable list, read the file paths one by one from the queue, execute each task, and then move on to the next. Implementing this should not be too difficult. May I ask if there is a scheduled plan for adding batch execution functionality to WhisperDesktop in the near future?

Hello, I am from China and my English level is not very good. This is a machine-translated article. I hope it can make you understand

mrdylanyin commented 1 year ago

Hi there, you can download cli.zip from release page and in powershell write shell script like this: .\main.exe -m ..\models\ggml-medium.bin (Get-ChildItem -Filter "*.wav")

canaskpw commented 1 year ago

May I ask how to convert the CLI to Chinese subtitles

.\main.exe -m ..\WhisperDesktop\ggml-medium.bin -osrt "DEMO.mp4"

jkfhto commented 9 months ago

Hi there, you can download cli.zip from release page and in powershell write shell script like this: .\main.exe -m ..\models\ggml-medium.bin (Get-ChildItem -Filter "*.wav")

‘.\main.exe -m ..\WhisperDesktop\ggml-medium.bin (Get-ChildItem -Recurse -Filter "*.mp4") -ovtt -t 1‘

Adding the '-Recurse' parameter results in the following error 'Unable to decode audio file "06. Automated Market Maker.mp4", MFCreateSourceReaderFromURL failed'