MoneyPrinterPlus previously used voice recognition services from various cloud vendors to work with video and subtitle recognition of speech.
But many of my buddies say that cloud services are unaffordable.
Well there's no way around it, MoneyPrinterPlus went live with the latest version that supports the fasterWhisper native speech recognition model.
Hurry up and experience it.
software preparation
Of course, the prerequisite is that you need to download the MoneyPrinterPlus software.
Download Address:/ddean2009/MoneyPrinterPlus
If you use it well, you may want to give a STAR to support it. After software version v4.1, MoneyPrinterPlus already supports the fasterWhisper native speech recognition model.
Installation of the masterWhipser model
The fasterWhipser service is called directly by MoneyPrinterPlus. So there is no need for a third party fasterWhisper service.
But we need to download the corresponding fasterWhipser model into MoneyPrinterPlus.
MasterWhisper model download address:/Systran
You can see that there are many kinds of models in it, you can download the corresponding models by yourself according to your needs.
How do I download it?
Go to the masterwhisper directory of MoneyPrinterPlus:
cd fasterwhisper
Execute the git clone command:
git clone /Systran/faster-whisper-tiny tiny
MoneyPrinterPlus currently supports the following model names:
'large-v3', 'large-v2', 'large-v1', 'distil-large-v3', 'distil-large-v2', 'medium', 'base', 'small', 'tiny'
So when you git clone, you need to rename the model directory in the faster-whisper repository to the name of the model supported by MoneyPrinterPlus.
For example, the model for faster-whisper-tiny is called tiny, so we renamed it when we git cloned it:
git clone /Systran/faster-whisper-tiny tiny
The above command creates a local directory for tiny. Inside the directory are all the model contents of the faster-whisper-tiny.
Configuring faster-whisper in MoneyPrinterPlus
We launch MoneyPrinterPlus.
In the basic configuration area:
Local speech recognition models can be configured.
model name is the name of the model you downloaded.
The device type can be cpu, cuda or auto.
The compute type supports 'int8', 'int8_float16', 'float16'.
Once configured, in the AI video area.
In the speech recognition configuration we select the local model, and we can use it for the masterWhisper.
Similarly in the video mixing area, we can select the local model to use the fasterWhisper.
summarize
Because it is a local run of fasterWhisper, there may be some problems with the environment in running it. You can refer to the instructions of fasterWhisper to solve them.
Click on me for more highlights.