Paul Boucher - Blog #AIPaul Boucher - Blog #AIhttps://www.paulboucher.com/blogs/tag/AIWed, 25 Mar 2026 18:18:12 -0700http://zoho.com/sites/<![CDATA[Stealing The Future]]>https://www.paulboucher.com/blogs/post/stealing-the-future

Some Video Game Producers Want To Steal The Future of Voice Actors

It used to be when a voice actor landed a contract with a video game company like Electronic Arts, Activision, or Disney, the results could be career-altering and, sometimes, even life-altering. 

Importantly, they were mutually beneficial, especially if the work was ongoing. The company had fantastic performances to build and sell a video game with (think Red Dead Redemption, Gears of War, Assassin’s Creed, and SO many others!). The actors were part of creative teams that helped create additional chapters in fantastic stories.

Now, these same companies want to screw over the voice actors whose performances helped them make billions. 

And that’s why SAG-AFTRA is striking games produced by Activision Productions Inc., Blindlight LLC, Disney Character Voices Inc., Electronic Arts Productions Inc., Formosa Interactive LLC, Insomniac Games Inc., Llama Productions LLC, Take 2 Productions Inc., VoiceWorks Productions Inc. and WB Games Inc.

You might remember the writers’ and subsequent actors’ strikes recently. The union fought for and won some protection from the unethical practices proposed by some of the companies. 

A simplified version of one issue for actors was that various producers wanted the right to use AI to digitally replicate a background performer without consent or compensation and deploy them in additional productions without paying the actor. 

What video game companies are proposing is the same unethical garbage the movie studios were: they want the unfettered right to digitally replicate any performer/performance WITHOUT compensating the actor after the initial performance has been captured.

Let’s call this what it is: it’s stealing the future of these voice actors’ careers. 

The industry standard arrangement currently holds that when a performer is called back in to add to a game, or recreate a character, they are compensated. The industry wants to use AI to create a digital replica or clone, and NOT pay the performer. Ever again.

Importantly, they don’t even want the performer to KNOW. They want permission to make actors say things they didn’t say, creating new “performances” from their cloned performances without their consent.

Here’s the much more polite SAG-AFTRA language explaining the points above:

“At the bargaining table, the employers have rebuffed our proposals and countered with loophole-filled language that negates any protections they claim to be offering. Employers understand that our members’ performances have value — that’s why they want to use them. It’s only fair that members be protected and paid for the value their performances create.

Among the more egregious loopholes, employers are only willing to obtain your consent and compensate you for the use of your “digital replica” if it is recognizable as you, which carves out nearly all movement work and the majority of voice work too, such as performing in character voices, voice matching and efforts. They want to reserve the right to have your digital replica say or do things that are religious, pejorative of a protected class or that endorse a political position without telling you and obtaining your agreement first. They also claim movement performers are not really performers, but rather data — even though they continue to hire experienced professional actors and stunt performers to do this work. They are using this language because they don’t want to give movement performers the consent, control and compensation they deserve. “

There is absolutely NO way that the people negotiating on the side of these companies don’t know that this is bordering on criminal. 

Whether you’re a union performer or not, the union is fighting the good fight in this one.

To read a comparison between what the union is proposing and the counter offer by the video game companies, go to this page.

The shine has been off the shiny new toy of AI for anyone working in a creative field for a long while. These are some of the reasons why. 

Know your rights. Understand the issues. Don’t let these companies steal your future.
]]>Thu, 22 Aug 2024 16:36:46 +0000<![CDATA[Keep It Real]]>https://www.paulboucher.com/blogs/post/keep-it-real

I’ve had clients tell me lately they’re having a tough time convincing their customers to use a human narrator rather than AI and/or text-to-speech voices, the shiny new toy in their production arsenal.

Those same clients are also NOT leveling with their customers that AI is often more expensive, time-consuming, and difficult than hiring a qualified human voice actor.


Here’s an example: when words can be pronounced differently depending on context. For example, “conduct” can be pronounced as it would be for a code of conduct. Alternately, there’s the word's pronunciation when you conduct an orchestra.

 

An AI or text-to-speech voice will get this wrong almost half the time, requiring time-consuming, manual fixes from a team member.

But those same clients are not divulging this manual-fixes-after-AI-screw-ups issue to their customers. Importantly, they are distorting the true cost of AI by notcharging their clients for the revision time. This is bad business practice. In addition, it materially contributes to many human creatives losing work by propagating the myth of AI “efficiency” in this context.

 

The result for those clients? Their customers are moving forward with the mistaken impression that AI is “faster,” “more efficient,” and “more cost-effective” than human narration.

 

As an experienced voice actor, I’m not blind to the fact that there will be many use cases where AI voice is the right tool for the job. For example, in a crisis like the wildfires in Jasper, it might be much more expedient to use a text-to-speech voice for phone lines, messages on websites, etc., with developments occurring so rapidly in real-time.

 

At the end of the day, though, one of the primary roles of a human voice actor is to infuse emotion, personality, and authenticity into narration, making it more engaging and relatable. AI voices can’t do that. AI hasn’t lived. It can “borrow” and “simulate” experience but will utterly fail to translate that into tangible human-felt emotion.

 

In simpler terms, an experienced human voice actor "gets it." AI voices can’t “get it” because they don’t care. There’s no human investment in your message. AI voices tend to sound flat and dead just when there should be emphasis.

 

I know of a middle management-level project manager within a large provincial government agency here in Alberta (Alberta Health Services) who has bluntly stated that they’ll never hire a human voice actor again.

 

When told a human actor might make concepts easier to digest and positively affect learner retention rates with clear, engaging, and engaged narration, that manager also bluntly stated that they don’t care about learner outcomes. Their obligation is purely regulatory. If the learners don’t take in the training, they say that’s on THEM.

 

Honestly, if ANY area of learning could use a human touch, it’s narration for Health Services.

 

Many producers are playing checkers, not chess, when considering the shiny new toy and could be better evaluating the use cases for human voices and skills versus AI or text-to-speech.

 

I’ve already replaced plenty of text-to-speech voices after a client was charged for the initial development. I’m happy to do it, but I wonder if any thought was given to how that customer experience might have been better if the use case had been properly analyzed and the right choice to use a human narrator had been made, to begin with. 

]]>Fri, 26 Jul 2024 17:49:54 +0000<![CDATA[A New Normal?]]>https://www.paulboucher.com/blogs/post/a-new-normal

"...Got to kick at the darkness 'til it bleeds daylight..." ~ Bruce Cockburn

File this under “you don’t get if you don’t ask,” or “there is light at the end of the tunnel.”

 

As a voice actor with a 25-year history of narrating bilingual eLearning, it's easy to discern a new normal in 2024.

 

More than a handful of eLearning clients have said the ugly part out loud: "We're having a tough time convincing clients to use a human narrator. They all want to try the shiny new toy: AI voice."

 

This particularly affects voice actors' bread-and-butter work, which is high-frequency, short-duration explainer and eLearning projects for actors living outside the big commercial production markets.

 

HOWEVER, when collaborating with true creatives, it's amazing what happens when the conversation becomes a win-win negotiation.

 

A localization company in the US that helps to convert videos into local versions (in my case, think US/Canadian English or Canadian French) approached me about collaborating in the near term and needed an agreement signed.

 

I got hung up on a passage in their agreement regarding Intellectual Property and sought clarification. In my request, I mentioned the concept that, as a voice actor, my voice is my intellectual property. Would the client be comfortable signing an agreement explicitly recognizing that, specifically in an AI context?

I proposed the legal language that the North American Voice Actors Collective (NAVA) posted on its website for the voice-acting community to use. You can find NAVA at https://navavoices.org. 

 

To my delight and surprise, this client created a written agreement with all five points verbatim and accompanied it with a message supporting the rights of creatives like voice actors to their intellectual property. The whole conversation surprised and delighted them as well.

 

I encourage you to have this conversation with any prospective client. It will help to ensure your voice acting career remains viable.

 

1.  Client expressly agrees not to utilize any portion of the Talent’s file, recording, or performance for purposes other than those specified in the initial Agreement between the parties, including but not limited to the creation of synthetic or “cloned” voices or for machine learning.

 

2.  Client shall not utilize any recording or performance of Talent to simulate the client’s voice or likeness, or to create any synthesized or “digital double” voice or likeness of Talent.

 

3.  Client specifically agrees not to sell or transfer ownership of any of the original files recording the performance of Talent to any third party for purposes of using the files for Artificial Intelligence, such as text to speech, or speech to speech uses, without Talent’s knowledge and consent.

 

4.  Client agrees not to enter into any agreements or contracts on behalf of Talent which utilize any part of the original files recording the performance of Talent for purposes of using the files for Artificial Intelligence, such as text to speech, or speech to speech uses, without Talent’s knowledge and consent.

 

5.  Client agrees to use good faith efforts to prevent any files of recordings or performances stored in digital format containing Talent’s voice or likeness from unauthorized access by third parties. If such files are stored in “the cloud,” Client agrees to utilize services that offer safeguards through encryption or other “up-to-date” technological means to prevent unauthorized third-party access.

 

Thanks for reading. Other thoughts on the voice acting industry are in other blog posts. Naturally, I hope you spend a few seconds listening to the voice (s) inside my head on the demos page!

]]>Wed, 10 Jul 2024 16:20:49 +0000<![CDATA[AI's Galloping Evolution And The Adaptation Challenge]]>https://www.paulboucher.com/blogs/post/ai-s-galloping-evolution-and-the-adaptation-challenge

So yesterday brought another iterative development that is genuinely another leap in the AI technology changing our lives, and livelihoods.

This is the latest version of ChatGPT, ChaptGPT 4o, the “o” in the name standing for Omni.


This is not an either/or reflection.


One of the most talked about new features is the new voice interaction within the software and how natural it can sound when interacting with a human. Remove the human interaction, and once again the AI voice can sound mechanical in short order.


To my eye, that single aspect continues to point the way forward for creatives of all stripes, although my context today is voice acting: “without human interaction…”


The way forward is a narrow bridge to a future where humans make themselves an indispensable enhancement to pure AI.


So, as always in a sustainable business for small vendors like individual voice actors, the challenge is to have a “value” conversation, as opposed to a “price” conversation.


For creatives of all stripes, how do you express the value you bring as the enhancement to AI being used? How does your contribution make the result exceptional, rather than momentarily startling?


How can you quantify that value? What makes your value a compelling business proposition? What does it bring to the table that the tech by itself cannot? How do you pitch it?


All this evolution has me thinking about these questions. I don’t have any definitive answers. I will be testing my thoughts in real-world situations with clients and continue to have conversations with them to learn more. I encourage other creatives to do the same.


My feelings are also informed by an interesting exchange with one of my eLearning clients last week.

He confirmed what we’ve been feeling and seeing in the industry. A lot of low value, low-stakes eLearning is being automated by AI voices. That’s not to mean that all of this learning, processed (it’s not being narrated, only humans do that) is of low-quality.


It used to be painfully obvious when this sort of learning landed. It was awful to listen to, and “off” in a way that became obvious after listening for short lengths of time.


As the technology gallops through its evolution, that’s changed.


This particular client said “out loud” what a lot of people on the voice acting side of the industry are fearing: that many clients are clamouring for “automation” rather than human choices on the narration side.


That’s due to a large number of factors, and although cost and profitability are high on the list, there are others. Speed. Simplicity. Consistency. All things in which the tech can easily notch wins against humans.


It’s also due to clients wanting to be perceived as being on the bleeding edge, so they will “overuse” the new toy, and there will be an inevitable “correction” as the technology finds its place.


The client also mentioned a new term – which led me further down this path to thinking that there’s a high-value role for humans in this specific field – “meta humans.”


For a meta human to exist, there must be a human base. At “this point” in the technology’s development, like many other AI attempts at “purely synthetic” creation, if something is purely based on a synthetic base, the quality of the creation degrades very quickly to a human eye or ear - from a human perspective.


In short, humans are essential to the equation. It’s up to us humans to make our case convincingly to the other humans.

]]>Tue, 14 May 2024 14:49:03 +0000<![CDATA[On The Subject of AI]]>https://www.paulboucher.com/blogs/post/on-the-subject-of-ai

It seems lately that every day provides an opportunity to consider AI and the ethics around it.

Last week, for example, I learned that a client who’s moved into an AI-related business was using a portion of a past French narration in a presentation demonstrating the capabilities of their AI to switch languages so a presentation could be instantly adapted and localized. 

The client neither asked my permission nor sought to compensate me for using my intellectual property – my voice – for a new use of a long-ago recording. 

This was one of two things: an unfortunate decision based on ignorance of the above or a deliberate decision based on the assumption that I’d likely never hear about it, and what was I going to do about it anyway? Understand that this client is using this narration to sell their technology, so essentially to profit from it without compensating the voice actors involved. Again, the benefit of the doubt allows that this is a whole new frontier, and it was done with no ill intent. However, independent of intent, theft is theft.

A couple of things crossed my mind:

1 - The successful legal battle that Canadian peer Bev Standing was forced into to have TikTok remove her voice from the platform and compensate her appropriately. They had acquired audio files from another company that had recorded Bev’s voice. That company sold the files to TikTok without Bev’s knowledge or consent.  The result was they used her voice, that voice actor’s intellectual property, to create new intellectual property for other clients without compensating her. The eventual settlement and resolution resulted from the hard work of a US attorney who rightly pointed out that the platform had illegally profited from using that actor’s intellectual property without due compensation for the unauthorized use of her voice.

2 - That directly led attorney Robert Sciglimpaglia Jr to work with an American industry association called NAVA to craft an AI rider that freelance voice actors could add to their contracts when working with clients.

It’s important to note here that most freelance voice actors working without a union contract go to great pains to vet their clients and essentially work on the principle of a virtual handshake, assuming that agreed-upon use terms, etc., will be respected.

So I’d like to share the AI rider, now attached to my terms of use, and used by many actors across North America and Europe. Please note the guidance to voice actors in red. The association added those notes to make actors aware that although these terms are designed to delineate the IP rights for the voice actor clearly, the rider, as a whole, is intended, with very few exceptions, to be part of a negotiation. So, there is flexibility in the terms to be adjusted on a case-by-case basis. 

PLEASE NOTE:

1 - This is not a “take it or leave it” Rider.  The client may have issues or want to change some of the language.  Changing language is normal for any contract.  If they want to change it and you don’t know what to do, or if you aren’t sure it’s already covered in their agreement, please contact NAVA, and we will refer you to an Attorney to review the rider and any requested changes.

ARTIFICIAL INTELLIGENCE RIDER

THIS RIDER is attached to the Agreement dated ___________________ between the parties _______________ (Talent) and __________________________ (Client) and is intended to replace and supersede any conflicting language in that Agreement.  (This sentence means that if something is different in the original agreement, then this agreement will control in Court.)

1 - Client expressly agrees not to utilize any portion of the Talent’s file, recording or performance of Talent for purposes other than those specified in the initial Agreement between the parties, including but not limited to creation of synthetic or “cloned” voices or for machine learning. 

2 - Specifically, Client shall not utilize any recording or performance of Talent to simulate client’s voice or likeness, or to create any synthesized or “digital double” voice or likeness of Talent. 

Most clients agree with the language in Paragraphs 1 and 2.  If they push back on these, then they may be planning on using the recordings for voice cloning or AI, so be sure to ask for more info about usage.

3 - Client specifically agrees not to sell or transfer ownership to all or part of any of the original files recording the performance of Talent to any third party for purposes of using the files for Artificial Intelligence, such as text to speech, or speech to speech uses, without Talent’s knowledge and consent.

4 - Client agrees not to enter into any agreements or contracts on behalf of Talent which utilizes all or any part of any of the original files recording the performance of Talent for purposes of using the files for Artificial Intelligence, such as text to speech, or speech to speech uses, without Talent’s knowledge and consent.

Some clients may think you are trying to restrict how they sell or use the end product after production with this paragraph, but that is not the intent.  Paragraph 4 says that they won’t sell or transfer your original recordings so that your voice can’t be cloned.  If the client wants to use the files to fix or add something in this particular job that they hired you for, that is not prevented by this paragraph.

5 - Client agrees to use good faith efforts to prevent any files of recordings or performances stored in digital format containing Talent’s voice or likeness from unauthorized access by third parties, and if such files are stored in “the cloud” Client agrees to utilize services that offer safeguards through encryption or other “up-to date” technological means from unauthorized third party access.

This is self explanatory, and it is not a be all end all in case a hacker gets a hold of your files.  It just says the client will use their best efforts to prevent that instead of doing nothing.

This rider is available to members and non-members of NAVA at this link: https://navavoices.org/synth-ai/ai-voice-actor-resources/#nava-synth-ai-rider. 

One of the recommended uses of this rider is to approach past clients and have an agreement in place to govern the use of past narrations. This would be particularly applicable to voice actors working in fields being cannibalized by AI, unethically or otherwise.

Off the top of my head, I can think of IVR/Telephony, eLearning companies and corporate narration/explainer video companies 

I will be following up with my client about that long ago narration. I expect a goodwill win-win resolution, which could include everything from a commitment not to use the narration anymore, or ideally, compensation for the current use, and a written commitment to compensate me for each and every single instance my voice is cloned and used for a new project. I’ll post an amendment to this post with the conclusion of that conversation at a later date.
]]>Mon, 22 Apr 2024 14:54:02 +0000<![CDATA[Localization!]]>https://www.paulboucher.com/blogs/post/localization

Bilingual content is often adapted or translated for another region or country for large corporations. 

The practice is called localization.

Localizing content to a specific region's culture, language, jargon, and values can significantly impact how the local learner receives and understands it. By tailoring the content to match the cultural norms and preferences of the region, it becomes more relatable and resonates better with the locals. Localization helps create a stronger connection between the company and the viewers.

Many low-value productions are now using AI to create text-to-speech narrations from one language to another.

As text-to-speech technology keeps improving, it's important to note that AI can only “simulate” human nuance. It doesn’t take long before an inflection or tone sounds “off”.

Unfortunately, that can leave a negative aftertaste about your brand in the viewer’s mind.

A human bilingual voice actor has inherent advantages when it comes to voice acting for localized content. Fluent in multiple languages (and usually, cultures), the voice actor grasps the nuances and subtleties of languages, including cultural references, idioms, and colloquialisms. This understanding ensures that the voice actor can effectively convey the source material's emotions, humor, and cultural context to the learners in the region's language, for example, US or Canadian English. They really can be different, eh!

Localization is not just about translating words; it also involves adapting the content to fit the cultural context. Using the correct or relevant cultural idioms demonstrates an understanding and respect for the local culture and helps build rapport and trust with the audience. It also avoids potential “misunderstandings.”

There have been numerous examples of translation mistakes that have hurt the reputation of companies or individuals. One famous example in advertising is the case of Pepsi's slogan "Come Alive with the Pepsi Generation," translated into Mandarin as "Pepsi brings your ancestors back from the grave." This translation error caused confusion and offended the Chinese audience by referencing a sensitive cultural topic. That mistake highlights the importance of hiring professional human translators and voice actors. They know cultural nuances and can help the client's instructional designers avoid those pitfalls. Don’t even get me started on machine translation services!

The long and short of it is this: localizing content to a specific region's culture, language, jargon, and values is crucial to create effective content, effective communication, and building connections. Human, bilingual voice actors with an understanding of multiple contexts have advantages in accurately conveying the intended meaning and emotions behind the content. Adapting the context and using the correct cultural idioms in translated or adapted content is vital for relatability and avoiding misunderstandings. Translation mistakes can significantly impact reputations, emphasizing the importance of professional localization and the expertise of voice actors as the "last stop" in bridging cultural gaps.
]]>Mon, 22 Jan 2024 01:02:33 +0000