New KDEnlive Release - How to use Speech to Text
Vložit
- čas přidán 11. 09. 2024
- #KDENLIVE #SpeechToText #NateCiraulo #Photolearningism
KDEnlive 21.04 is out with some new bells and whistles. Let's check out how to use Speech to Text for auto subtitle generation!
DOWNLOAD KDENLIVE
kdenlive.org/e...
VISIT CREATOR TOOLBOX - FREE SEARCH TOOL
www.photolearn...
FREE CZcams THUMBNAIL 50 PACK
bit.ly/PLThankY...
BUY MY CONTENT CREATOR EBOOK ($0.99)
amzn.to/2xsuHZh
ENHANCE CZcams CONTENT MANAGEMENT:
TubeBuddy = www.Tubebuddy....
VidIQ = vidiq.com/#_l_ma
CONNECT WITH ME!
Instagram = / photolearningism
Twitter = / nateciraulo
WEB = www.NateCiraulo...
FREE Search Tool = www.photolearn...
Disclaimers: All opinions are my own. Links in the description are typically affiliate links that let you help support the channel at no extra cost.
Left a comment and leaving another one just to say you are awesome, loved the bloopers :)
Thank you so much, Vuyani Moyo! Really appreciate the kind words and support! 😀
That's awesome!!!!! Thanks!!! It's realy helped me!!!
You wonder why it stopped in the short distance, you have to grab that speech ribbon and stretch it the full length of your video, and make sure your edit subtitle tool is not checked under "project" "surtitles" , make sure and drop down to "speech recognition" and check it.
Great video, is it available for Ubuntu users?
Hi Lara Lima! Thank you so much for watching! I've had the opportunity to test within Ubuntu in the latest #KDEnlive release (21.12) and confirmed this does work just as reliably there too! For Linux, it is necessary to add Python separately, then add the PIP SRT library (pip install srt - If Python is missing, Terminal will show the sudo command to install it first). Hope that helps!
Hey, is Kdenlive Speech-To-Text BETTER than Subtitle Edit Speech-To-Text??? I'm fully aware that they BOTH use Tesseract and these Vosk Models...but I'm curious if the ACTUAL recognition of Speech (SPECIFICALLY IN MOVIES) is better than Subtitle Edit?? Most of the Transcription Software I've checked out only seems to pick up "Interview/Single-Speaker" style of Speech with NO BACKROUND Music or Chatter.
I'm searching everywhere how to edit the style of those subtitles and aside tagging everything in HTML, which is quite the hassle, I don't know how to do it.
Kdenlive is awesome, if I have a clip that is presented by say someone with a french accent is there a voice model pack that is customised for that?
Hi Vuyani! Thank you for the question! I took a quick look over language models (alphacephei.com/vosk/models) and thus far, they look to focus on language structure rather than annunciation. Your question is an interesting case - my best suggest would be to use a model in desired language and test for accuracy. Another approach is to try out the "Kaldi toolkit", which appears on the model page under the "Training your own model" section. I haven't tried this out myself, but from the description, it may help to create and refine a language model, teaching it to account for accent. Hope that helps!
Thanks for your prompt response, this will definitely come in handy.
Thanks.
Nearly one year after you made this video, but maybe you still see my comment? thank you for your tutorials, they help me out getting started with kdenlive, my fery first attemp in videocut sw. But with speech to text I have problems, I followed along all steps, on Win 10, have python 3.10.4, installed by path, current kdenlive 21.12.3 ( I tried as well the standalone like the other one), I tried on 2 Computers.....and set up all the other stuff importing the speechmodels . Kdenlive seems to start to process, and after few seconds it says, it has imported subtitels and I get no error. But I never get any subtitle !?! all configuration checks are without errors... so, if you have an idea, what I can do still?
You have to place the play head in the expandable (stretchable) blue band on the speech to text track.
@@genghis2510 thx you, I had done this... the problem was not showing to me, that there was still a prob in missing something in python. after reinstalling, it showed up to me, its missing sth, but I could fix it now.
it dosent work for me
i have tried everything to install but fail
i installed python,i tried to run that .bat file and fail
it just won't work.how the hell am i going to know what should i do when i never try this thing before
i translate english video to chinese and i really need to get this done so i wont have to type while i listening to the video
its a pain.help
When installing Python 3.9 you need to check also the "Add Python to PATH" thing that shows up just after clicking on the python installer.If you check that and install python 3.9 the .bat file will work. I wanted to add that If version 21.04.3 does not work for you try downloading version 21.04.0 STANDALONE download.kde.org/stable/kdenlive/21.04/windows/ that one works for me until they will release stable version 21.08 soon.
Poor speech-to-text can't recognize the name of your channel, and no wonder. It is even written incorrectly (missing an N) in your links shown in the video. :-D
Anyway, to see that screen you mention close to the beginning you should click on View and Text Edit, at least in version 21.08.1. Not particularly intuitive, but so it goes. I have not succeeded in transferring the text generated from that window to the subtitle though. I'll try doing this the way you did. Fingers crossed.
Thanks J. Alves! I laughed when I connected the dots on this one - not sure how I missed it but glad to have it fixed going forward!
surprising they use python
I'm on Linux. I get this error when I try to generate the speech to text: ERROR (VoskAPI:Input():kaldi-io.cc:756) Error opening input stream conf/final.mdl
terminate called after throwing an instance of 'kaldi::KaldiFatalError'
what(): kaldi::KaldiFatalError
Any idea what I need to do to fix it? I can't find anything on the net.
Also, how can you add another VOSK? I'm not able to add another one, or to remove the one that's producing the error and replace it with another one. I'm stuck :(
UPDATE: I got it to work with another vosk :)
Great! Glad you got it to work!