Synthetic Narration?

R

Richard Crowley

Veteran
So what is the state of the art for synthetic, computer-generated speech/narration?
I ask because I stumbled across this "documentary" on YouTube today and I was curious about the VO narration.
The phonemes, words, and phrases definitely sound machine-generated, yet the inflection is quite human.
I can't tell whether it is very good (perhaps human tweaked) machine-generated speech,
or maybe a recording of a human narrator which has been diddled to sound vaguely artificial?

https://youtu.be/UOKnRjAWjh4

I have no opinion (at least no positivie opinion) of the content or the subject matter.
It seems like typical YouTube conspiracy twaddle to me. But the narration intrigues me.
 
I wondered if it might be old-school speech synthesis, i.e. a real person is recorded speaking thousands of words for a basic vocabulary, then someone (or a program) stitches the required words together in sequence to create narration.
 
That sounds like the standard speech synthesis that is used on all those spoof cartoon like videos about the film and TV industry.

You know the ones where the indie producer is hiring a soundie and he is using the RED as it is THE BEST camera or the one where the DOP has to use a techno crane rathe than the SUPER techno crane.

No doubt someone thought it would be a cheap way of narrating their doco as all you do is type the words in and it spits out the speech.

Sadly some bright spark with fob this off as "The New Techno Voice Over Style" and it will become de facto for all those Yoo Toob feature docos shot with the "searching for focus style" and the "wobbly DSLR style" with location interviews and audio courtesy of the 5.1 RODE Video mic surround that is plonked on top of the rig!

Then numerous students will re-create this new style by hiring name voice overs and putting them through a Melodyne pitch corrector to grade their voices and emulate the classic tones!

Am I getting too old or cynical for this industry ???? :undecided
 
No Gary, you're on target.

As a "senior" member of SAG-AFTRA, and one who has done tons of narration, the whole table shifted after the 2000 strike.

That, several economic crashes and a few other misfortunes have unfortunately made "good enough" an industry standard.

Many of the younger people wrangling VO talent don't even have a clue as to what a true narration pro can bring to the party.

They often can't hear it in their heads. They'd rather be doing anything than running a narration record session.

To them, an auto-bot (you've got mail!) delivery might actually be perceived as "something new" to play with.

I took a piece of copy that I performed as a radio spot in 2009 and spent a few hours to make the following. I programmed in the angles, edits, eye moves, head tilts and arm movements. Here's what happened….

https://www.youtube.com/watch?v=wLszi3dnA_o

Regards,

Ty Ford
 
Last edited:
The poster of that conspiracy video said in his comments that he was not a native English speaker. Perhaps using the synthesized voice was preferable to trying to decipher a heavily accented narration by the author, considering that he probably didn't have the money to hire a professional English-speaking narrator either.

At the very least, it sounds a lot better than the weird Indian-accented female voice they use for our local NOAA weather warnings nowadays.

- Greg
 
An interesting discussion over on EEVblog Forum about the VO narration of a product video.
The script was written in technically dodgy Chinglish, but the narrator(?) pulled it off without laughing out loud.
Although it was clear that they didn't know the difference between milliAmps and MegaAmps! :-)
And there continues to be debate about whether this was a human VO actor, or whether this was synthisized.

http://www.eevblog.com/forum/testgear/voice-over-artist-reading-owon-marketing-chinglish/

 
I vote for synthesized. A non-technical VO actor might not have caught the "mega-amps" gaffe, but almost certainly would have corrected some of the missing articles and distorted grammar almost spontaneously. The levels and intonation are also more consistent throughout the video than a live actor would have been likely to deliver. It's a very good synthesis program, though, and is just demonstrating the GIGO principle. If it had been fed a well written script, nobody would have bothered to listen to it that closely, and we might not even be having this conversation.

Hey, that scope they're advertising looks pretty nice, too. Miles ahead of the 40-year-old HP I have on my workbench!

- Greg
 
You must log in or register to reply here.
Back
Top