itch.io

Hi tbruinsma, thanks for the nice comments!

Speaker Diarization is a frequently requested feature, and I will add it in the next update of the application. Hope that helps!

Hi Nicos, I am the developer of this application.

This application is based on a deep learning attention model which basically either works or not.

With the provided demo you can test with your images if it's going to be helpful or not with your own images.

Ah, I see that you mentioned that you bought this about a week ago. Maybe you should have used the demo first.

In any case, I'm guessing you will be asking for a refund, if that's the case, here is all you need to know to do that: https://itch.io/t/129454/how-are-refunds-handled

Hi Wad Mabbit Society, happy to hear you liked this software.

At the moment, it can only transcribe media files, not directly in real time from the microphone.

One option you can do is to record the sounds coming from your system, and then feed that recording into the software. This will get you the best quality, as real time transcription requires a simpler model, and is also not currently planned for the short term at least.

Hi Rebork5555, you can keep the transcription in Spanish by selecting "Keep original language" in the Settings menu.. You can see that option by going to the Edit menu, then clicking on Settings. You'll see something like this:

Hi rebork5555, you can download a demo of the program to see how it works. Just click on the Download button next to PrivateTranscriberPro DEMO v1.5.2.

You are absolutely correct Kijkeenolifant, the current version only uses CPU. In a future version there will be an option to accelerate it with your GPU.

The point was to first make it available to everyone, and then making it better over time with new releases. Anyone that buys it will have access to future versions anyway, forever!

I ended up discovering a different way of uploading files (butler) which is much nicer and doesn't have the 1GB restriction. So, as promised, I just updated the app to include five different model sizes, which are now available in v1.5.2.

I didn't want to include all the 3 versions of the large model as the download is already at about 5GB so I only added large-v1 which seems to be the one with least amount of issues in general, but if you want to use any of the other 2 large models (v2 or v3), you can simply copy the model you want to use to the models folder and change its name to large-v1 and it will use that model instead when you select the most accurate setting.

Thanks Kijkeenolifant, great to hear you liked it.

What you say is a valid point. I originally planned to include all the models in the application, but ended up with a file that is way larger than itch.io's maximum allowed download file size (1GB), so I ended up including only a subset of them to make the application pass this constraint.

Having said that, it looks like I can manually request itch.io for a larger maximum file size, so I will update this tool if they increase this limit.

Hi sild928,

Yes, the software saves an output image with exactly the same resolution of the original.

There is no limitation on how large the image can be, as long as the operating system is able to load it.

Hi robinne,

This happens because Apple blocks applications from unidentified developers. The same will happen with the full version.

If you want to run this application, you can read the official guide from Apple here: https://support.apple.com/en-us/102445#openanyway

Hi Christine,

I just published Depth And Normal Maps from Image which might be relevant to what you are looking for.

Hi Christine,

Yes, it is possible to generate both, the depth map and the normal map from a single image.

I might end up publishing a tool like that at some point.