Şaban Özdemir, Media PR Unit Manager of Üsküdar Üniversitesi Corporate Communications Department and Voice Actor, evaluated the highly debated artificial intelligence technologies that have reached a level where they can imitate real human voices.
AI-based applications are finding their place in every aspect of life
Özdemir, stating that AI-based applications are finding their place in every aspect of life, emphasized that it would be correct to define artificial intelligence, which brings risks as well as making life easier, as 'the digital atomic bomb of our age.'
Özdemir noted that this neutral technology can be used for both good and bad. “If we go back just 20 years, how many of us could have imagined that the technology that merely read written texts would, by today, not only read these texts but also articulate and voice them in a lively and impressive manner? We can easily say that on digital platforms, in content production, technology always surprises and amazes us compared to the previous day. We always lag behind technology, struggling to keep up with its pace… We can say that we are experiencing the same level of advancement, change, and transformation in the field of sound as well.” he said.
Can texts be converted into lively, emotional voices?
Özdemir reminded that there are dozens of AI-based voice generator platforms. “You can mostly benefit from these various platforms through paid subscriptions or memberships, utilizing services for reading and voicing written texts. We can see that texts can be converted into lively, emotional voices. These AI-based voice generator platforms can also be chosen with multiple language options. Yes, these platforms offer the ability to quickly convert text into high-quality sound with instant voice creation capability. Moreover, they even provide the option to select alternative voices from a sound library,” he informed.
AI-supported voice generation is unsettling in scope…
Şaban Özdemir, stating that AI-supported voice generators offer a remarkable AI tool for production, continued:
“We would not be wrong if we said that the point it has reached is at times unsettling. Indeed, it is clear that AI's voice generation, cloning, and replication are causing concern among professional voice actors who earn their living specifically in this field. Considering it on a global scale, there are thousands of artists who derive their livelihood from this.
So much so that the initiative taken by the American Federation of Television and Radio Artists SAG-AFTRA in recent months, figuratively speaking, brought voice artists to the streets. The announcement of a groundbreaking agreement with an AI voice production studio to produce copies of artists' voices drew strong reactions from professional voice artists, who criticized this initiative with actions and strikes.”
Zeki Müren voiced the Parla anthem with artificial intelligence
Özdemir pointed out that informally operating AI-supported voice generators are gaining unfair profits by cloning the voices of professional voice artists. “Artists who do voiceovers for commercials, films, video games, and books have also witnessed their voices being cloned through artificial intelligence and used in various content without their consent. This is labor theft… This is an observable situation in our country, as it is on a global scale. Not just for voice actors… It can be observed that the voices of some political leaders are also imitated using artificial intelligence technology, made to sing songs and anthems, and subjected to propaganda. One example is the beloved artist Zeki Müren, who passed away in 1996, being made to voice the 'Parla' anthem with artificial intelligence. As you may recall, it became one of the most shared and viewed videos on platforms as soon as it was posted.” he said.
Exploitation of voice actors…
Şaban Özdemir, stating that the reckless use of synthesized and cloned voices in every field (audiobooks, podcasts, documentaries, dramas...) leads to the exploitation of artists and a confrontational legal situation, shared a personal experience on this matter with the following sentences:
“A few years ago, someone asked me to do a voiceover for a few pages. They informed me that after my voiceover, my voice would be copied and used in voiceovers for different areas. A voice demo would be taken from me, my voice would be taught, introduced, and copied using deep learning models for AI technology. This way, every subsequent text entered would be voiced using the defined voice. Of course, I did not accept this…”
Is 100% human voice possible?
Özdemir said, “AI-based voice generator platforms have not yet been able to fully replicate a real human voice with 100% naturalness and realism, especially in terms of intonation, emphasis, and accent. Texts can indeed be converted into human-like voices in seconds, but are they realistic? No, they are not! It's still a mechanical voiceover, devoid of emotional tone…”
Özdemir concluded his words by stating that in addition to legal regulations, an AI-based technology capable of detecting cloned voices could be developed to prevent unofficial AI voice cloning. “Legal protection is essential to prevent unauthorized use of voice artists' voices. This technology, which distinguishes AI voices from human voices, will provide a serious deterrent for fraudsters.”
