View previous topic :: View next topic |
Author |
Message |
Bruce Boardmeister

Joined: 06 Jun 2005 Posts: 7977 Location: Portland, OR
|
Posted: Thu May 19, 2022 3:12 pm Post subject: Something To Chew On, Then Spit Out? |
|
|
Since A.I. was brought up here I thought this article clipping might be of interest. It's from the Kiplinger Newsletter, a major source of financial/investment information. It's always intriguing and quite frustrating when a writer writes about an industry they've only partially researched. A lot of assumptions are made and spewed out as fact. None the less, here's what's being touted about our potential digital opponent to some less than savvy investors. As always, your thoughts are welcome!
B _________________ VO-BB Member #31 Enlisted June, 2005
I'm not a Zoo, but over the years I've played one on radio/TV. . |
|
Back to top |
|
 |
todd ellis A Zillion

Joined: 02 Jan 2007 Posts: 10529 Location: little egypt
|
Posted: Thu May 19, 2022 4:16 pm Post subject: |
|
|
admittedly, i am often difficult to use - but i've never gotten $10k out of an audio book either. _________________ "i know philip banks": todd ellis
who's/on/1st?
 |
|
Back to top |
|
 |
Bruce Boardmeister

Joined: 06 Jun 2005 Posts: 7977 Location: Portland, OR
|
Posted: Thu May 19, 2022 4:56 pm Post subject: |
|
|
With the exception of narration by major personalities, no one pays $10K for a standard length audiobook read. Plus with an AI book you still need QC, editing, mastering, and correction of misread words.
I’d like to hear one of these fancy AI programs read something with the pacing and nuance of a human being. They can’t be there yet. Remember the movie Blade Runner with the replicants performing everyday jobs? You could tell, and it was creepy.
B _________________ VO-BB Member #31 Enlisted June, 2005
I'm not a Zoo, but over the years I've played one on radio/TV. . |
|
Back to top |
|
 |
Philip Banks Je Ne Sais Quoi

Joined: 20 Jun 2005 Posts: 11075 Location: Portgordon, Scotland
|
Posted: Fri May 20, 2022 2:00 am Post subject: |
|
|
AI teaching me to fly? It's a utility and it will work fine.
AI telling me about the joy of the first solo flight?
Crtl +Shift + type M E C will make the AI voice sound like it is making eye contact with the listener. Not happening.
The reason that most voiceover people are seeing a downturn in workload because of AI is due to the fact they are selling their voices as a utility not a personality.
When AI cares or has a director comment "Wow! You can hear the twinkle in the eyes" then and only then I done for! |
|
Back to top |
|
 |
Bruce Boardmeister

Joined: 06 Jun 2005 Posts: 7977 Location: Portland, OR
|
Posted: Fri May 20, 2022 1:04 pm Post subject: |
|
|
So, I thought I'd dig a bit deeper into each of the companies the author of the newsletter clip mentioned but didn't explore:
DeepZen.com is a pure translation from digital print to AI audio service. I listened to their "Voices". "Todd's" fiction demo had an oddly paced description of a character that sounded more like a grocery list than colorful fiction. The big surprise? Their narrator "Edward's" fiction demo was clearly, to me, Edward Hermann's voice STOLEN and REPURPOSED. His read was not as wooden but was still annoyingly irregular. How much is their service? If you want the full treatment as you might get with a standard audiobook production house, it's $129 per finished hour, plus a $69 fee to set up some form of distribution, and a $25 annual fee for continued distribution.
Descript is not an audiobook narration company. They offer an online recording service for people who want to make their own videos such a podcasts or informational videos. They also turn spoken words into text.
Pozotron is a weird one. It's an AI service that just performs Quality Control on the audio you send them and sends you a report. They also offer pronunciation help. But that's it.
Scribe Audio offers no samples, no rates, just a bit of puffery about their abilities. Speaking of puffery, here's the copy on the first page of their website:
Quote: | The Gutenberg of Audiobook Production is Here.
Get your backlist converted to audio for as low as 1% of studio cost. Hollywood class voice actors, crisp high fidelity audio and turnaround in days, not weeks. |
I've written to ask for samples and with general interest. We'll see.
Lastly, Speechki. The name is probably derived from the home city of this company, Omsk in Siberian Russia. No good samples or clear pricing in the website, and if you want any information they insist you have a 30 minute phone conversation with one of their staff. Uh, no.
With what samples I could hear, I believe these services are for buyers who think this inferior (compared to professional narrators) service is GOOD ENOUGH for their customers.
B _________________ VO-BB Member #31 Enlisted June, 2005
I'm not a Zoo, but over the years I've played one on radio/TV. . |
|
Back to top |
|
 |
Tim Bick

Joined: 28 Oct 2021 Posts: 11 Location: Brighton, UK
|
Posted: Wed May 25, 2022 11:25 am Post subject: |
|
|
Does AI know when and for how long to pause? Does AI know how to drag the length of a word out? Can it accentuate specific words in a specific way? And a million other tiny human things? This kind of reminds me of when all the drummers (I used to be one) panicked about the rise of the drum machine. Yes it replaced a specific sector of work, but to get a really good drum track you still need a good drummer to programme the thing. |
|
Back to top |
|
 |
Philip Banks Je Ne Sais Quoi

Joined: 20 Jun 2005 Posts: 11075 Location: Portgordon, Scotland
|
Posted: Wed May 25, 2022 11:33 am Post subject: |
|
|
AND don't get me started on mastering the mystery that is the Iranian Double-Pause! |
|
Back to top |
|
 |
Tim Bick

Joined: 28 Oct 2021 Posts: 11 Location: Brighton, UK
|
Posted: Wed May 25, 2022 11:45 am Post subject: |
|
|
I’ve heard it done and still didn’t quite comprehend what I had heard. Extraordinary. |
|
Back to top |
|
 |
Bruce Boardmeister

Joined: 06 Jun 2005 Posts: 7977 Location: Portland, OR
|
Posted: Wed May 25, 2022 11:52 am Post subject: |
|
|
I'm still trying to master the Iranian Double Pause. So farsi so goodsi.
B _________________ VO-BB Member #31 Enlisted June, 2005
I'm not a Zoo, but over the years I've played one on radio/TV. . |
|
Back to top |
|
 |
Moosevoice Backstage Pass
Joined: 16 Nov 2012 Posts: 437 Location: Iowa
|
Posted: Thu May 26, 2022 9:34 am Post subject: |
|
|
'bleep, blorp.' $10k please!
Seriously though.
I remember doing morning radio and what our consultant would tell us, everytime, is that CONNECTING with people, having personality, being relatable to your audience is what winning shows do.
Doing a vapid 'morning zoo' or prank calls doesn't translate to long-term success. Being a human does.
Obviously that doesn't slide right over to Voice acting but I think the point is still valid.
People will always want to be moved by words and the right actors delivering those words bringing the 'feels'. The AI will have it's place and not all jobs will totally need a human to spit into the microphone for 'em but there's space for both.
And when they DO use AI, it'll be a lot like the 'Uncanny Valley' with CGI graphics in movies and TV shows. They'll hear the AI voice but something won't quite seem right to them either consciously or otherwise. And THAT will affect the message trying to be delivered as the end-user will here something that's not quite sitting right with them and completely miss the emotive reason of the message.
Words! _________________ www.moosevoice.com |
|
Back to top |
|
 |
todd ellis A Zillion

Joined: 02 Jan 2007 Posts: 10529 Location: little egypt
|
Posted: Thu May 26, 2022 10:37 am Post subject: |
|
|
Talimogene Laherparepvec, Empagliflozin, Ustekinumab ...
go for it AI. _________________ "i know philip banks": todd ellis
who's/on/1st?
 |
|
Back to top |
|
 |
Kim Fuller DC

Joined: 29 Jan 2011 Posts: 641 Location: Portlandish, Oregon
|
Posted: Thu May 26, 2022 3:20 pm Post subject: |
|
|
Ustekinumab. I've got a little PTSD from all the times I said that in a pharma Important Safety Information video. |
|
Back to top |
|
 |
Bruce Boardmeister

Joined: 06 Jun 2005 Posts: 7977 Location: Portland, OR
|
Posted: Thu May 26, 2022 3:53 pm Post subject: |
|
|
I hate to say it, but I tried those four drug names with IBM's Watson computer, and it did a pretty good job. I didn't research the proper pronunciations, but it sounded very reasonable to me. Of course this is the computer that has defeated world chess champions.
B _________________ VO-BB Member #31 Enlisted June, 2005
I'm not a Zoo, but over the years I've played one on radio/TV. . |
|
Back to top |
|
 |
todd ellis A Zillion

Joined: 02 Jan 2007 Posts: 10529 Location: little egypt
|
Posted: Fri May 27, 2022 6:52 am Post subject: |
|
|
it's not so much the words themselves (as you know) it's using them in a sentence - in context - and making it sound like you know what they mean (even when you don't)
i'm sure Ai will get there - but - hopefully - i'll be dead by then. _________________ "i know philip banks": todd ellis
who's/on/1st?
 |
|
Back to top |
|
 |
|