Can OpenAI Whisper do this? I see Punjabi on the language list. I use it through Faster Whisper within SubtitleEdit, output SRT files and then load them into VEGAS.

https://github.com/SubtitleEdit/subtitleedit/releases

thanks for suggestion

let me check once

@iEmby

As far back in late 2022, I made a script to add Speech to text in Vegas using OpenAI whisper (and a procedure on how to install whisper, and supporting stuff), not sure if the procedure still works, as in the mean time easier install methods and further development of whisper is done.

Here is the link to the post:

https://www.vegascreativesoftware.info/us/forum/speech-to-text-via-whisper-openai--137928/#ca863411

I do not know if you have the happy OTTER tools for Vegas, as they also have speech to text using whisper with Vegas I believe, although I have not used that part of the happy otter stuff yet.

@bitman thanks for response..
i tried your script it open dialoge box of buttosn but when i tried to add subtitles in project
it shows
 

System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. ---> System.IO.FileNotFoundException: Could not find file 'C:\Users\iamem\Downloads\Telegram Desktop\video_2024-08-19_10-17-44.mp4.srt'.
   at System.IO.__Error.WinIOError(Int32 errorCode, String maybeFullPath)
   at System.IO.FileStream.Init(String path, FileMode mode, FileAccess access, Int32 rights, Boolean useRights, FileShare share, Int32 bufferSize, FileOptions options, SECURITY_ATTRIBUTES secAttrs, String msgPath, Boolean bFromProxy, Boolean useLongPath, Boolean checkHost)
   at System.IO.FileStream..ctor(String path, FileMode mode, FileAccess access, FileShare share, Int32 bufferSize, FileOptions options, String msgPath, Boolean bFromProxy, Boolean useLongPath, Boolean checkHost)
   at System.IO.StreamReader..ctor(String path, Encoding encoding, Boolean detectEncodingFromByteOrderMarks, Int32 bufferSize, Boolean checkHost)
   at System.IO.StreamReader..ctor(String path, Encoding encoding)
   at EntryPoint.MakeLinkedList(Vegas myVegas, String myPathPlusFileName)
   at EntryPoint.FromVegas(Vegas vegas)
   --- End of inner exception stack trace ---
   at System.RuntimeMethodHandle.InvokeMethod(Object target, Object[] arguments, Signature sig, Boolean constructor)
   at System.Reflection.RuntimeMethodInfo.UnsafeInvokeInternal(Object obj, Object[] parameters, Object[] arguments)
   at System.Reflection.RuntimeMethodInfo.Invoke(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)
   at ScriptPortal.Vegas.ScriptHost.ScriptManager.Run(Assembly asm, String className, String methodName)
   at ScriptPortal.Vegas.ScriptHost.RunScript(Boolean fCompileOnly)

@iEmby, here's the actual error:

Could not find file 'C:\Users\iamem\Downloads\Telegram Desktop\video_2024-08-19_10-17-44.mp4.srt'.

@iEmby, here's the actual error:

Could not find file 'C:\Users\iamem\Downloads\Telegram Desktop\video_2024-08-19_10-17-44.mp4.srt'.

 

yes sir but why it is not creating srt file? even it show' Subtitle saved in text file

@iEmby I tried my script on Vegas 22 b93 and it still works, make sure you use the script version v8, unfortunately this forum does not seem to be able to re-edit older post as it once was possible, even if the post was originated by yourself. Version V8 can be downloaded by a link in in the same post, but towards the end.

If you have difficulties with the script it usually has something to do with file paths and/or audio file location and environment variables not properly set or not having enough user permissions.

System.IO.FileNotFoundException: Could not find file 'G:\Books To Brain\Short Stories\7 Time is God (10 mint)\Time is God (Edit).veg.srt'.
   at System.IO.__Error.WinIOError(Int32 errorCode, String maybeFullPath)
   at System.IO.FileStream.Init(String path, FileMode mode, FileAccess access, Int32 rights, Boolean useRights, FileShare share, Int32 bufferSize, FileOptions options, SECURITY_ATTRIBUTES secAttrs, String msgPath, Boolean bFromProxy, Boolean useLongPath, Boolean checkHost)
   at System.IO.FileStream..ctor(String path, FileMode mode, FileAccess access, FileShare share, Int32 bufferSize, FileOptions options, String msgPath, Boolean bFromProxy, Boolean useLongPath, Boolean checkHost)
   at System.IO.StreamReader..ctor(String path, Encoding encoding, Boolean detectEncodingFromByteOrderMarks, Int32 bufferSize, Boolean checkHost)
   at System.IO.StreamReader..ctor(String path, Encoding encoding)
   at EntryPoint.MakeLinkedList(Vegas myVegas, String myPathPlusFileName)
   at EntryPoint.FromVegas(Vegas vegas)
   --- End of inner exception stack trace ---
   at System.RuntimeMethodHandle.InvokeMethod(Object target, Object[] arguments, Signature sig, Boolean constructor)
   at System.Reflection.RuntimeMethodInfo.UnsafeInvokeInternal(Object obj, Object[] parameters, Object[] arguments)
   at System.Reflection.RuntimeMethodInfo.Invoke(Object obj, BindingFlags invokeAttr, Binder binder, Object[] parameters, CultureInfo culture)
   at ScriptPortal.Vegas.ScriptHost.ScriptManager.Run(Assembly asm, String className, String methodName)
   at ScriptPortal.Vegas.ScriptHost.RunScript(Boolean fCompileOnly)

still same issue..
i guess it is not creating srt file.

@iEmby I tried my script on Vegas 22 b93 and it still works, make sure you use the script version v8, unfortunately this forum does not seem to be able to re-edit older post as it once was possible, even if the post was originated by yourself. Version V8 can be downloaded by a link in in the same post, but towards the end.

If you have difficulties with the script it usually has something to do with file paths and/or audio file location and environment variables not properly set or not having enough user permissions.

@bitman PM either myself or another mod with the changes you want and we should be able to change it for you.

@bitman sir... am i missing something?
i have wisper installed in my PC already.

is there any need to add its path in script? which i did'nt do

SubtitleEdit is way easier to setup if you just want to see how well it works with Punjabi.

@iEmby 

@bitman sir... am i missing something?
i have wisper installed in my PC already.

is there any need to add its path in script? which i did'nt do

No, this is not needed, my script will store both .srt and .txt in the same folder as the source audio, just make sure the folder has full user permission and make sure you have read the word file on installing everything, especially regarding possible issues on environment variables not properly set.

As a matter of fact you can test out whisper itself without the Vegas script. This has to work before you attempt to use the script, the same word document explains how to test whisper to establish if whisper itself is working standalone, on the audio files outside of the script.

I would consider moving the file OUT of the "downloads" folder...

SubtitleEdit is way easier to setup if you just want to see how well it works with Punjabi.

Sir.. i tried SubtitleEdit. it is not so accurate and also taking so much time as English subtitles developed in just 10 seconds.

even i downloaded there all large models.

still there is no accuracy. and even it does't develop all subtitles.

In that case I wouldn't expect Whisper to work any better with Punjabi even if you could get it to work within VEGAS. Not all languages are equally supported as the amount of info to train with on the web in English is so much greater than others.

In that case I wouldn't expect Whisper to work any better with Punjabi even if you could get it to work within VEGAS. Not all languages are equally supported as the amount of info to train with on the web in English is so much greater than others.

yes i agreed... even if i fix that script error too then also whisper can't do it.

see this how much time whisper take and it gives only some lines from multiple lines and all those are not accurate.

i tried whisper for my english tutorials it works awesome but not for Punjabi.

but i have one question.

how WhatsApp or other message app can generate 95% correct subtitles for Punjabi or any other language through Gboard (Google Keyboard) Voice Input.

look how i have to do it.

can we use Gboard's technology in the above script instead of whisper?

One thing I'd suggest large v2 over v3. It might work better. Whisper works entirely locally so it's not surprising that it's slower than offloading it to a powerful server. It's actually fast... but only if you have a NVIDIA GPU (CUDA). I can transcribe an hour in a few minutes on my GPU.

One thing I'd suggest large v2 over v3. It might work better. Whisper works entirely locally so it's not surprising that it's slower than offloading it to a powerful server. It's actually fast... but only if you have a NVIDIA GPU (CUDA). I can transcribe an hour in a few minutes on my GPU.

ok sir.. let me try once.,,
and whats your opinion about this

can we use Gboard's technology in the above script instead of whisper?

@iEmby if Vegas speech to text isn't working for your need subtitle edit would seem to be next best choice. That large model is working much too slowly for you, maybe that's because it can't all fit into your 4GB Vram or it can't use AMD GPU's(check task manger, is it even using your GPU), could test with a smaller model.

Also I think that CPP engine is for CPU, and you have a good CPU, maybe that's faster but you likely wouldn't want to use the large model as it will be too slow. As for accuracy, if you can read these green subtitles you could tell me

As for accuracy, if you can read these green subtitles you could tell me

0% accurate sir..

because it make subtitles of Hindi (national language of INDIA) but not what i want Punjabi (language spoken in a state of INDIA). I also get same results by large v2. Only Punjabi comes with large v3 but not accurate and not complete. Hindi is already available in VEGAS Speech To Text.

this is Punjabi subtitles i created manually as i stated at top of this thread.

@iEmby That's no good at all, and the problem with LargeV3 is that it's less accurate than V2 in many tests, whisperX looks to do better with V3, but that would be another technical challenge to get that working with subtitle edit and may not be worth it in the end if it also is so slow on your computer, the wer% is word accuracy the smaller the better, the cer% is character accuracy , smaller the better

sir... can you tell me

from where i can get whisperx.exe ??