I'm pinning this to the top of the forum page. Please provide any feedback about this feature that you can as you use it. We want to get a good read on how people are using this so we can improve it for the general release in VEGAS Pro 22.

Here I leave my comments, I hope you take them into account.

 1) Please leave the option to save the favorite language to be transcribed.  Most people only speak one language and choosing it manually is tedious.

 2) Please add a button or script to remove all silences.  If there are 100 silences, going manually makes no sense.

 3) Add the option for the words to appear at the same time they are mentioned.  Capcut transcribes sentences but have a option for the words appear in order as the word is mentioned.

 4) Add more animations for the texts, one or 2 pop or gamer type animations would be very useful

 Thank!

I'm leaving the screenshot report here from one user in Facebook group:

https://www.facebook.com/groups/13702281564/posts/10161614911126565/

The second screenshot seems to be about text to speech, not text-based editing.

Hmm, puzzled. I use 'speech to text' lots to auto generate subtitles (e.g. for social clips ready for facebook). In the new version, it seems very hard to do this. Previously I could select the clips I wanted transcribed, choose 'speech to text', click Analyse and it would generate subtitles for just the portion of those clips used on the EDL.

Now even if I select just the clips I want, it gives me a long list of all (some? doesn't look like all, but lots of clips anyway) the clips on my EDL. I scroll through and find just the two clips I want to transcribe, click analyse and it takes ages to analyse the entire clip (not just the portion showing in the EDL). Then I go to 'subtitles beta' and click generate, and it puts subtitles that don't exactly match the edited clips on the EDL. They are close, but they start and stop too late in the clip (as in the words at the start are missing), as if it's trying to sync to the audio but picking the wrong in and out point. I always had to manually correct the auto-subtitles for layout, but now there's a load to correct to add back the missing words.

I think it also ignored half the formatting options I selected - so the font is correct, but they're not on a double line and line length isn't what I chose.

Not sure this is a step forward for what I need to do. is there are a way to go back to the non beta version? Thanks.

I'm on 365 pro subscription.

I also just want to generate subtitles from speech, like in previous versions. Is there some option to select only what is on the dialog or voice over track?

I'm pinning this to the top of the forum page. Please provide any feedback about this feature that you can as you use it. We want to get a good read on how people are using this so we can improve it for the general release in VEGAS Pro 22.

If that's the case I'll elaborate a bit more on what I'd want from a "track transcribe" feature.
1. If it works in a similar way to the current AI transcribe then having the tracks in a separate list from the media items would be very helpful.
2. Ideally it could be even easier in a right click menu for example: Select Track(s) to Transcribe -> Right Click -> Speech to Text.
3. Adding a right click menu option for this on media items in the timeline would also be very helpful.
4. The ability to transcribe multiple tracks/media items at once through this right click feature would also be very helpful.

I look forward to future developments on this feature. The act of finding the part of an hours long video where I said something would be made a lot easier with this.

Trying out the transcription to text based editing beta feature.

It shows silent/pause areas, but you can hide them.

You can click on the Transcript window and it adjusts the timeline marker.

Suggestions/bugs:

• I'd prefer if transcript generation were a local process instead of cloud based. Privacy considerations and I'd wonder if a local process would be faster while still as accurate or even better depending.
• The interface isn't clear you need to do the transcript first because it lets you access the editing option but it's a blank screen.
• Having auto-ripple on clears out all of the text in the transcript window after deleting anything, so it appears to be broken.
• Some sections of silence are including spoken words. For example in the screenshot the word "6" is attached to the silent portion rather than before or after.
• You can't do the reverse of clicking on the timeline to select text in the transcript editing window.
• It would make sense to have this transcript window part of the left panel tabs.
• I don't see a natural way to play the audio while using the text editor to confirm it is selecting what it needs to do.
• Deleting individual pieces or groups of text might not be positioned properly so some way of dealing with that from an interface standpoint would be nice.
• I'd also like it to work on timeline audio rather than source clips.

I like this idea! I could eventually see it faster to use than timeline editing for certain types of videos.

I would like to see in addition to removing all silence over x amount, the ability to leave a buffer of silence on either side of the cut/split.

Here's a use case. The speaker has a pause as they are parsing out how they say what they want to say to conclude a sentence. They then resume and move along with the dialogue. There still needs to be a small amount of silence left to make the dialogue flow naturally.

Suggestion

There should be an option to transcribe all clips in the timeline.  In my case I usually add two hours of material and if I could delete all the silences with one click it would be extremely useful.  Currently I have only been able to do it by nesting the entire timeline and analyzing the entire project.

Why "Text to speech" and "Speech to text" not working - "The service is unavailable"?

@Sergey-Kulikov good question. It isn't working here either.

I'll check back later this evening.

I wish we could have the old "speech to text" function back as a separate function. It worked very well for the way I work with subtitles. The only way for me to use this new function now is to have just one audio file in the project and to make sure there are no edits done to it. It then takes a very long time to analyze the file. When I choose to have the subtitle shown on two lines it still delivers one long line. The subtitles are not "timed" to the speech as they were in the previous version. All the "text plates" come in a long row without any space between them. So now It takes much more time for me to finish the subtitling.

I think a search option should be given in Text Based Editing where user can search any word or line split out it from text. 

beccause some times there can be long interview or podcast videos. 

If possible new language request. as other Indian languages are already there, please add one more which is Punjabi (Gurmukhi) language.

Are there any new youtube videos out on these features? Some of those that were making tutorial videos on youtube just haven't uploaded much in a long time.

This is a beta feature, so there aren't any tutorials out for it just yet.

i think i found something strange while using speech to text.

i just import an video and then save it as a nest project.

then i import that nest project into new project.

then i run speech to text on nest. it is behaving strange. processing percentage cross 1000 numbers.

and processing bar is already full.

but as it is working in background, when you will cancel by clicking on cross then it will show you text.

if it is a bug then it should be note down.

one more thing i got to notice.

when i click on subtitles it show me lines but lines are not of equal length.

when i change words per line then it too change only uper lines.

try to reproduce it.

may be it is because of para starting from that line. so it is not processing subtitles settings on lines of new para.

Murray Bacon

I have had ongoing problems with V20 and V21. During V20 I put in a fault request and never heard back that it had been addressed. I just waited for V21 and crossed my fingers. Some problems fixed in V21 and many more problems added. Is it worthwhile to put in a fault request?

If it hasn't already been covered in this thread add your concern here?

https://www.vegascreativesoftware.info/us/forum/vegas-pro-build-300-text-based-editing-beta-feedback--145247/#ca916668

did anybody try to reproduce it. or i am doing something wrong. please tell me i have similar work ahead.