Open axayjain opened 4 years ago
Are you sure you're not talking about the "Disfluencies" option? This addes uhs and ums and errs that weren't in the original transcript. I suspect if you left it blank or added 1 word or a "space" character it might produce the results you were seeing?
You can see here how it's done with curl (programatically) https://github.com/lowerquality/gentle/issues/256
gentle is for alignment. So I think you should do speech recognize first to obtain the transcript.
See the answers that people gave me when I asked this question some time ago.
Using the command line is much too complicated for me to understand. I use Docker Desktop to run Gentle's 'Docker container' and use Kitematic to give Docker a graphical user interface.
Since I could not get a straight answer as to how to install the full language model to the official Gentle Docker container, I installed a Gentle Docker container with the full language model already installed.
the online demo version can decode phonemes without the transcript. But the code doesn't work without it. Is there any way to do that in code too?