85 lines
5.0 KiB
Plaintext
85 lines
5.0 KiB
Plaintext
|
|
Episode: 3731
|
||
|
|
Title: HPR3731: Speech recognition in Kdenlive
|
||
|
|
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr3731/hpr3731.mp3
|
||
|
|
Transcribed: 2025-10-25 04:44:33
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
This is Hacker Public Radio Episode 3,731 from Monday the 21st of November 2022.
|
||
|
|
Today's show is entitled, Speech Recognition in KD in Live.
|
||
|
|
It is the 10th show of DNT and is about 6 minutes long.
|
||
|
|
It carries a clean flag.
|
||
|
|
The summary is a brief description of how you can use speech recognition to transcribe
|
||
|
|
your clips in KD in Live.
|
||
|
|
Hello and welcome to another exciting episode of Hacker Public Radio.
|
||
|
|
This is your host DNT.
|
||
|
|
I want to share something I just found out about a KDE in Live.
|
||
|
|
So this is a nonlinear video editor application from a part of KDE, I guess.
|
||
|
|
I used this a very long time ago.
|
||
|
|
The first time I had a computer with a Linux operating system installed, it was Ubuntu
|
||
|
|
back just before the Unity desktop came out.
|
||
|
|
I think it was maybe version 8 or something like that.
|
||
|
|
I had it installed on a little network and it was so bad that KDE in Live, I don't
|
||
|
|
know why, I'm sure it was for some specific reason but I didn't know it back then.
|
||
|
|
So KDE in Live couldn't even preview the video in the application itself.
|
||
|
|
So I cut the first videos that I ever cut on that even though I couldn't see the video
|
||
|
|
I could only hear the sound.
|
||
|
|
So what I would do is I would play the video on VLC and I would kind of take some audio
|
||
|
|
cues from that and then in KDE in Live based on the audio cues I would cut the video.
|
||
|
|
So I cut a couple of videos like that and then I kind of decided that I was going to pursue
|
||
|
|
this for work for a while video and then of course, you know, soon I moved into, I kind
|
||
|
|
of went into the idea that now you have to have an Apple computer to do this kind of work
|
||
|
|
so I eventually bought a Mac and then I used Final Cut Pro for a long time and then now
|
||
|
|
I don't work doing this anymore so I don't need to use this kind of thing.
|
||
|
|
Nobody will care what I use so now I've been using KDE in Live again and I only do video
|
||
|
|
for my own things just for fun.
|
||
|
|
But anyway, so using KDE in Live to cut a little video recently it was an interview and
|
||
|
|
then I found out it now has this thing, it's like a speech recognition and it can basically
|
||
|
|
transcribe your interview for you.
|
||
|
|
You have to install a couple of Python libraries which the KDE in Live does for you.
|
||
|
|
I assume it runs PIP to do that.
|
||
|
|
I don't know, it doesn't really tell you.
|
||
|
|
But anyway, it installs those two things for you in one click and then you have to go
|
||
|
|
to a website which the application itself links you to that website then you download
|
||
|
|
a model.
|
||
|
|
There are multiple different models and there are models for several languages and the
|
||
|
|
models are like pre-trained models for speech recognition and then you load this model
|
||
|
|
into KDE in Live and you use it to do speech recognition on a clip in your project in
|
||
|
|
KDE in Live.
|
||
|
|
So then it'll generate this little window thing in KDE in Live that has the transcription
|
||
|
|
of your video and you can even click through and it'll take you to the clip and so then
|
||
|
|
you can literally click a word and then press the space bar and you'll start to see the
|
||
|
|
video right at that point where that word was said.
|
||
|
|
And it's really great.
|
||
|
|
I don't think there were any errors with the transcription.
|
||
|
|
I think everything was correct.
|
||
|
|
So this is really impressive because if you've ever done this before, you may know the
|
||
|
|
transcribing interview is kind of a big pain.
|
||
|
|
Nobody wants to do it.
|
||
|
|
It actually makes me think of some old interviews that I still have that I shot many years ago
|
||
|
|
and I never got to finish them or anything.
|
||
|
|
I could at least run them through this thing and get a transcription of the interviews
|
||
|
|
for at least for archival purposes if nothing else.
|
||
|
|
So really impressive stuff, just amazing to think that this is free software or open source
|
||
|
|
software perhaps I'm not sure.
|
||
|
|
But anyway, something that's freely available and that works very well.
|
||
|
|
And also that same application that I was using about 12 years ago when I first started
|
||
|
|
editing videos is still an excellent application on Linux to edit video.
|
||
|
|
So yeah, I came across this and I thought it might be of interest to hackers.
|
||
|
|
Recently we had an episode from MrX about editing videos as well.
|
||
|
|
He was using shot cut.
|
||
|
|
I've tried that myself but I found KDE in live was a bit better for me but not too different
|
||
|
|
anyway.
|
||
|
|
So and yeah, it was funny MrX was kind of in a pretty similar situation so that one
|
||
|
|
I was in about 12 years ago editing video on a computer and barely able to play the
|
||
|
|
video I was trying to edit and having to use VLC to assist me in knowing how to cut it.
|
||
|
|
Pretty cool stuff.
|
||
|
|
So anyway, thank you for listening.
|
||
|
|
Remember we're running a bit low on shows right now so if you can sit down and just talk
|
||
|
|
about anything that you've come across recently like I just did and then post it to hacker
|
||
|
|
public radio, we would really appreciate it because I like to have those shows coming
|
||
|
|
in every day even if they're just short and simple like this.
|
||
|
|
That's great because then I have more time to do other stuff.
|
||
|
|
So thank you and I'll see you next time.
|
||
|
|
An attribution 4.0 International License.
|