Camtasia 9 speech to text / captions

  • 1
  • Question
  • Updated 11 hours ago
Looking for help on speech to text options to create a file for closed captioning / video subtitles. I located a blog on the Camtasia main site that indicated there was a built-in option but I can't seem to find it. Any suggestions and/or other tools that you recommend looking at? I am hoping to not have to transcribe from scratch, but rather just clean up an auto-generated text file. 
Using Camtasia for Windows (current release) on Windows 10. 
Photo of Nadia Munro

Nadia Munro

  • 64 Posts
  • 10 Reply Likes

Posted 10 months ago

  • 1
Photo of Rick Stone

Rick Stone

  • 4479 Posts
  • 1997 Reply Likes
Hi there

Perhaps the link below helps?

http://www.screencast.com/t/YXeZc0uqG

Cheers... Rick :)
Photo of Nadia Munro

Nadia Munro

  • 64 Posts
  • 10 Reply Likes
Thank you for this. Just seems like a ton of work. Was hoping for something more automated similar to youtube's speech recognition tool (but where I could in fact generate a full transcript without having to do it manually). 
Thanks for the video.
Photo of rupert.fallows

rupert.fallows

  • 4 Posts
  • 1 Reply Like
Azure media services might help here
Photo of Nadia Munro

Nadia Munro

  • 64 Posts
  • 10 Reply Likes
thanks, I'll take a look
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
you can generate full project captions using speech-to-text; a single step can do it all, but you'll need to edit/correct the resulting captions for errors; the more you use STT on your system, the more accurate the results

I do STT over short sections of the project; gives me much better results

to add to Rick's suggestion, use this menu to access more speech to text tutorials ...

Camtasia Tutorials Menu

filter in the TOC for "speech to text"

item 08-12 might be most helpful
(Edited)
Photo of Nadia Munro

Nadia Munro

  • 64 Posts
  • 10 Reply Likes
Got it, I did locate the speech to text option and the voice training. Not sure why I was missing it before. Thanks.
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
I don't recommend Camtasia speech to text. It's extremely inaccurate and will cost you a lot of time correcting mistakes.It uses the Windows speech recognition engine and it won't learn your voice if you train it till you drop.

Dragon Naturally Speaking is the best way to go.It the tool I use. However, it's not cheap. https://www.nuance.com/dragon.html

There's a free plug-in for Microsoft WORD, Power Point and Outlook.

 It's a Microsoft Garage Project. I've played around with it and it's fairly accurate.  Much better than the Windows version.If you have Word, it might help you out.http://dictate.ms/

 
    Regards,Joe
(Edited)
Photo of Nadia Munro

Nadia Munro

  • 64 Posts
  • 10 Reply Likes
Thanks for this Joe, I will take a look at the options you suggested. My experience (only one so far) with the Camtasia built-in tool was not terrible... but any improvement/time savings is always welcome. Thanks again.
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
I use Dragon Naturally Speaking also and it's pretty good out of the box but,
the more you train it (use it) the better it gets.
That's pretty much the norm with most dictate software. Some never learn :(

Downside... as Joe states,  Dragons not cheap.

:)
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
I sent a one page E-mail to my nephew the other day using Dragon and Word. I used a  online word counter to count the number of words. There were 766 words. Dragon made about 10 errors in total.I didn't count them.Dragons performance wasn't on my mind.But that's an accuracy of around 99% or so.

Windows Speech Recognition is 70% accurate if your very lucky.Microsoft should hang their head in shame for offering it as a tool to aid the handicapped. I don't care if it is free.
In Camtasia, you spend so much time  looking for and correcting mistakes it's almost laughable.
 
kayakman,  you know so much about Camtasia. Your a huge asset to this forum. So don't take this wrong.
I'm using your tutorial as an example to demonstrate just how bad speech to text is when using Camtasia to do it. In this video  http://www.screencast.com/t/Udc5Xpk6PkD

You show 3 small captions being created through speech to text.Two are riddled with errors. One has only one error.
I counted up the words for all captions.
31 original words in total were created.
13 of them had to be deleted and replaced.
That equates to an accuracy level of below 60%. 

That's nothing short of "Are you kidding me?" in my opinion. I tried like heck to make the Windows 10 version of speech to text do a decent job last summer.Just to see if they improved it since windows 7. They haven't.

Among others, Audacity is revered by many as the go to tool for audio when using Camtasia.
Why? Because Camtasia's audio tools just aren't robust enough.

Windows speech recognition is horrendously inaccurate.

Members should be warned just how bad it is up front. A disclaimer if you will. Then point out alternatives.
Because less than 60% accuracy is unacceptable.

Case in point. If I were paying someone  $25.00 an hour for weeks on end. To waste their time creating and correcting captions in this inefficient way.
 Then realized I could have purchased Dragon Naturally Speaking. It would have payed for itself.
My employee could be further ahead in his/her workload as well.That, would tick me off.

I rest my case, this isn't a personal attack.Just common sense.IMO     

Regards,Joe  
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
no offence taken:)

I did a quick test this AM; got about 83% accuracy

https://www.screencast.com/t/yvpISwoCxx

I've used the CS STT tools since CS6; the CS6 tools yielded extremely accurate results; sometimes, 100% across many captions; in those days, it was a different operating system [Win XP and Vista]; the STT engine was much different; you could place the cursor into a blank bubble callout and execute a audio segment STT and it would type in the callout text directly in the callout; worked great with near-perfect results; I made 450 videos using that technique; I miss that capability

I've found that audio quality makes a huge difference; hence using noise removal before STT; but if I do many back-to-back Camtasia STT actions, the accuracy can often go north of 90%; accuracy seems to slip if it is a very long time between using the function?
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
83% is not a good result.

Overall, Microsoft's speech to text engine is terrible.

Dragon runs at 95%+  accuracy on a bad day if you have a cold.
Try downloading Dragon for a 30 day free trial. You won't ever want to go back to where you are now.

To be impressed with 90% accuracy, tells me you haven't used a good program to date.

Try anything but what your doing now. It's worth the trip.83% is horrible.That's technology from the dark ages.

Regards,Joe
Photo of rcarranza

rcarranza

  • 1 Post
  • 0 Reply Likes
I just did a test and about 30% accurate....almost every single sentence had errors....can't use camtasia:-(
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
did you "train" your STT ?

training can make a huge difference in accuracy

there are 3rd party STT tools, but they can be expensive; there is even a google speech to text app, but I've never tried it
Photo of rupert.fallows

rupert.fallows

  • 4 Posts
  • 1 Reply Like
Azure Media is the professional service from Microsoft and not the cheap Windows solution.
Photo of ccmark100

ccmark100

  • 2 Posts
  • 3 Reply Likes
Joe,  I have Dragon and Microsoft Speech to Text.  They both are way behind the power curve.     Google has been in development of Speech to Text and is at the next level.      There is no "Learning/Training" the application with your voice.  In fact, you can have a table of people talking one at a time and the recognition factor is great.  Again no training.     Go to this link and install the application.  You will be greatly surprised.          https://tinyurl.com/y7r8yvtw

I even install the Gmail Speech to text.  No more typing my emails either.
Photo of Joe Morgan

Joe Morgan

  • 5220 Posts
  • 2718 Reply Likes
I will certainly give it a try, If it's better than Dragon "Goodby Dragon"

I'm no fan of antiquated software.

Thanks {:> )
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
Here's my follow up. I tried Google's Voice to Text and it's not for me.

Voice to Text is fairly accurate I'll give it credit for that.
However, I read a few different things into the website.It made quite a few mistakes.

I use Dragon Naturally Speaking 15. It's works much better for me.

Plus, you're forced to use the Googles program online. I can use Dragon offline and in several programs.

It's much better than Windows Speech to Text. But Windows is scraping the bottom of the barrel.So it's not much of an endorsement.
But a great option for those using Windows.

Overall, Voice to Text did a pretty good job.

However, the program needs to learn your voice like Dragon does.

Dragon does Accuracy tuning. You teach it words based on your voice.You correct it when it's wrong. It compiles all this information and gets better all the time.
 
My current profile contains almost 1GB of information.
There's a lot going on under the hood when I open that profile.


Regards,Joe
(Edited)
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
Hey kayakman,

I made a  speech to text video for you as well.

It's a quick demo of Dragon Naturally Speaking functioning at over 99%  accuracy.

I used the same text you did in your video using Windows Speech to Text above.
However, I read an extra paragraph to emphasize how much better Dragon is by comparison.

 Your results came in at 83% accuracy.  https://www.screencast.com/t/yvpISwoCxx



(Edited)
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
Joe, thanks for the Dragon video; I'd never seen it used before; I've been aware for years that it it a better tool than what comes with Windows; but as you said, it's not a free-be 

the Camtasia/Windows combo give granular control over creating captions from a screencasted clip; I assume you've used Dragon in that capacity?; if so, how does that work flow go?
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
No I haven't kayakman, I don't normally use captions .

If I did. I could export the narration as a wav file. Dragon can transcribe voice recordings.

 I exported the same narration I created for the video from Camtasia as a test. I transcribed it with Dragon.I've been aware of the feature but I've never actually used it until today.

It yielded the exact same results as what you saw in the video.

You can copy to clipboard, paste it all into Camtasia Captions at once.

Or notepad or something along that line.Bring them in as a more targeted copy and paste approach.


 
 
Or you can do it piece meal through the Dragon Dictation Box. You just open the Dictation Box and talk to it.Transferring the captions one at a time.

I made a video to show that technique.

 
(Edited)
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
I have Dragon 12 and I can use it right in the caption field in Camtasia.

I don't need to use the dragon dictation box or what not.

Just click "Add Caption" speak my text.

click Add Caption" speak my text.
click Add Caption" speak my text.

and so on.

As Joe says you can also export your narration and use Dragon to transcribe it.




:)
(Edited)
Photo of Joe Morgan

Joe Morgan

  • 5629 Posts
  • 2920 Reply Likes
I don't think what you want to do is possible kayakman, Camtasia's direct interaction with Microsoft's Speech to text is what you're looking for.
It would be like trying  to hack a plug in if you ask me.

I was trying to something similar by making Dragons microphone input my speakers output by default.Then selecting a small area of narration on the timeline and playing it back.

 I needed to have that Dictation Box running in the background just to have control of the timelines playback. The playback controls were blocked by the captions window. So it's very clunky  at best. It wasen't worth the effort. Maybe there's another way?But  the Camtasia editor wasn't designed to interact with Dragon at that level and either was Dragon.

 Exporting the audio and transcribing it makes the most sense to me. Copying and pasting correctly spelled dialogs should be more efficient than correcting inaccurate Speech to Text all day.

What I know is this, when I use Dragon I get 98% to 99% accuracy on a regular basis.

I would think you can take advantage of this and make it work for you. If you actually can may be another story.
That's what 30 day free trials are for.
(Edited)
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
@ kayakman

I don't believe so.

Dragon dictation sources are limited to a microphone or recording device.
That's in Dragon 12 anyways.

I have not ever tried use with a recording device

There may be a workaround such as putting your mic by a speaker but only not practical, you could have issues such as feedback or bad accuracy results.

I would say the best method in your use example would be to export the audio and transcribe it with Dragon,  import it back in to Camtasia and apply the caption splits
where needed.

:)
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
Ya beat me to it Joe.

But we're on the same page.
Photo of Joe Morgan

Joe Morgan

  • 5629 Posts
  • 2920 Reply Likes
Pretty much, but using the microphone input as your speakers output gives you no distortion or outside noise. It actually works pretty good.
But it created to many other problems interacting with Camtasia.

 If you want to pursue it further, refine the technique.There's probably a way. I don't have the time to invest in this any further right now.

Regards,Joe
(Edited)
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
I too have no reason to pursue it farther.

I think in conclusion the best method is to transcribe the audio.

Time wise to export >transcribe >make corrections (if/where needed) >import>edit timing
would be the most logical and productive method.

:)
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
thanks all; but I've been looking at their web site; see offers for 30-day money back guarantee; no free 30 day trial?  am I missing something?
Photo of Dubie

Dubie

  • 950 Posts
  • 730 Reply Likes
I don't see a free trial anywhere either.

Best I could find on the site money wise was the Home v13 addition on sale. 30 Day money back

https://www.nuance.com/dragon/dragon-for-pc/home-edition.html

:)
(Edited)
Photo of kayakman

kayakman, Champion

  • 6241 Posts
  • 1828 Reply Likes
thanks again guys; we're all seeing the same; if I could get it to behave with the workflow I described, it might be worth it

I now only intermittently require STT for captions; most of my captions work is now either for project/video navigation, or video "key word" searching; those uses requires typing them in
Photo of Bob Lewis

Bob Lewis

  • 348 Posts
  • 172 Reply Likes
Would it be possible for Camtasia to have an option to use Dragon rather than Windows speech to text?

I would vote for that.
(Edited)
Photo of Joe Morgan

Joe Morgan

  • 5630 Posts
  • 2922 Reply Likes
I don't see how they could without some form of agreement with Nuance. "Makers of Dragon" They would need to design Camtasia to interact with Dragon as well.That would take some serious makeover I'm sure.
I think your better off setting your sites on lower hanging fruit.
Regards,Joe
Photo of Bob Lewis

Bob Lewis

  • 348 Posts
  • 172 Reply Likes
Well, I am definitely Not holding my breath.