AI has wiped out another industry

In the past two years, AI voiceover has swept across the entire industry at an astonishing speed, from advertising to gaming, from audiobooks to short dramas.

It is faster, more obedient, and also cheaper.

Those who originally relied on "voice for a living" have to redefine their value in the wave of AI.

Clients Are Almost Gone

With only a few days left for sample delivery, documentary director Helen found herself stuck with voiceover issues.

In March of this year, she was producing a biopic for a senior figure in the educational drama field. At the final stage of the project, she needed to condense a 45-minute film into a five-minute highlight.

The elderly man in the film is over eighty years old, and every frame is infused with decades of stage experience. Each line of narration must be concise and steady to support his entire life.

Helen reached out to several voiceover companies, but the samples she received left her frustrated. Either the voices were too young, lacking the depth of age, or the emotions were too exaggerated, disrupting the elderly man's calm demeanor.

"Good voiceover should blend into the visuals, but what they provided felt like two separate entities." She listened to the samples repeatedly, growing more anxious.

As the deadline approached, the editor reminded her, "Why not try AI?"

To her surprise, the effect generated by AI amazed Helen. The pauses, emphasis, and emotional fluctuations were just right, even closer to the film's essence than the previous human recordings.

She even suspected that the production team might be "using real people pretending to be AI."

Having been in the industry for over ten years, Helen has worked as a program director and has also shot commercials. She has always believed that voiceover determines the breathing rhythm of the visuals and is where emotions settle. In the past, the voiceover teams she collaborated with charged varying fees, from 100 yuan to 1,000 yuan per minute, with senior voice actors even charging by the second. Now, AI achieves almost the same emotional delicacy at an extremely low cost.

Helen began to worry, "Will those at the bottom of the industry be the first to be squeezed out?"

This concern became a reality for 28-year-old part-time voice actor Chengzi.

Chengzi first encountered voiceover in his second year of high school. At that time, an app called Voice Show was popular, and he would often sneak out to the old square in town with his phone when no one was home. In the empty echo, he mimicked the voices of different characters, sometimes a passionate anime protagonist, sometimes a gentle neighbor boy, as if creating a group of "invisible friends" for himself.

His favorite was the scene in "Fox Spirit Matchmaker" where Dongfang Yuechu saves Wang Quanfu. He recorded that line more than a dozen times before he was willing to send it out. During evening study sessions, he felt a little light in his heart every time he saw a "like" pop up on his phone.

He truly started earning money from voiceover during the summer of his sophomore year. Chengzi found a post in a Douban part-time group offering "1 yuan for 100 words" voiceover work and accepted it without much thought. Over one summer, he recorded dozens of movie commentaries for Baijiahao, earning over 300 yuan, which bought him his first microphone.

After graduation, Chengzi joined a trading company in Fuzhou. By day, he was a silent newcomer in the workplace, and by night, he returned to his rented room to record various movie commentary scripts into the microphone. That was his most relaxing moment of the day. To reduce echo, he was afraid to stick sound-absorbing cotton, so he piled several stuffed dolls in front of the desk Over the years, his voice-over price has risen from 1 yuan per hundred words to 5 yuan, and his equipment has been continuously upgraded. At the peak, he could have more than twenty clients actively reaching out in a month.

The turning point occurred in 2023. A blogger he had collaborated with for over a year stopped updating movie commentary, and another long-term partner directly switched to AI voices. During that time, he increasingly heard uniform, mechanical, and emotionally flat male voices while scrolling through short videos; they mostly had strange breathing patterns and lacked emotion, yet they "worked."

At first, he could tell immediately that it wasn't human voice. But this year, he found that he could no longer distinguish it with 100% certainty, and almost all the clients who used to reach out to him had disappeared.

The industry is sinking like a tide, and he stands right next to the line that is gradually being submerged.

Industry data shows that by 2024, the market size of China's AI voice and semantics has risen to 14.93 billion yuan, with a compound annual growth rate of over 28%. The mainstream voice synthesis platforms on the market can now provide thousands of voice types, covering various dialects and sub-styles.

45-year-old John has been a voice-over director in Guangdong for over twenty years. He first heard AI voice-over ten years ago. The voice at that time sounded somewhat stiff, and he didn't pay much attention, feeling it was still far from real voice-over work.

However, changes had already taken root in the shadows. After 2010, AI voice began to proliferate in scenarios requiring a large amount of standardized content, such as navigation and customer service. By 2024, 90% of in-car navigation systems, 80% of mobile map applications, and over half of online education platforms had adopted AI voice.

By 2025, the proportion of AI-generated audiobooks exceeded 40%. Clarity, uniformity, and pleasantness are the core demands of these products, and AI can complete a chapter recording in just a few minutes at almost zero cost.

In contrast, the chain for human voice-over is lengthy. Recording studios, directors, actors, and post-production—all steps require time and expenses.

In Guangdong, although the past market size was not as large as that of Beijing and Shanghai, it had always been stable. Cantonese dramas, children's animations, and audiobooks have long relied on human voices, with a single character sometimes accompanying a voice actor for a lifetime.

In recent years, John has clearly felt the foundation loosening. Fewer and fewer of his peers in the audiobook industry are receiving narration work, and new projects almost exclusively use AI, retaining human voices only for character dialogues.

Fields that emphasize "speed" and "cheapness," such as short videos, foreign language short dramas, and local advertisements, have also been rapidly occupied by AI.

John has to admit that AI is systematically taking over the most basic, stable, and sustainable jobs in the industry.

Voices Stolen by AI

While taking away jobs, AI is also rewriting the floor price of the entire industry.

Voice actress Xie Wanwan has been in the industry for 4 years, starting from taking freelance jobs and recording advertisements, gradually building a team of over 200 people. The team includes not only voice actors but also planners, post-production staff, illustrators, and screenwriters, forming a comprehensive studio capable of independently producing complete content.

Since the first half of this year, Xie Wanwan has clearly felt that the market is being squeezed thinner and thinner by AI The most obvious change is the collapse of prices, especially in the audiobook sector. Three to four years ago, high-quality voice actors could earn 600-800 yuan/hour, but now, some jobs have dropped to as low as 50 yuan/hour.

Colleagues have privately formed a "lowest price alliance," trying to band together to maintain a "100 yuan bottom line," but Xie Wanwan understands that the clients will not be soft-hearted and will directly hand the projects over to AI. Even the opportunity for auditions is gradually being taken away.

If the decline in prices is the overt confrontation, then "voices being cloned" is like a hidden arrow.

Last month, Junjun finished work and lay in bed scrolling through Xiaohongshu when her phone suddenly buzzed with a series of WeChat messages: "Have you done this voice?" "Quick, watch the video!" "This voice sounds like you!" She clicked on the link and after listening for just two or three seconds, she was stunned—this was her voice, but it was not her performance.

Her voice had been cloned by AI.

Junjun, from Hunan, graduated with a degree in broadcasting and hosting. Her voice is lazy, clean, and natural, perfectly fitting the type of "relaxed yet high-quality" voice that is currently favored in the advertising industry. From major ads for Nayuki and JD.com to short promotional voiceovers for street vendors, her voice can be heard everywhere.

After graduating in 2022, she worked as an editor at a dubbing studio in Changsha, learning the order-taking process for commercial voiceovers while editing. After returning to her hometown, she became an online commercial voice actor. Her parents supported her career by converting a room in their home into a dubbing studio and equipping it with a high-end microphone and a mobile recording booth.

Although she is a freelancer, her life is even more regular than that of office workers. She sits in front of her computer at nine o'clock every day waiting for audition requests, trying out twenty to thirty lines a day, and finishing work promptly at five-thirty in the evening.

Image | Junjun's dubbing studio at home

Three years ago, she laughed when she first heard AI voiceovers: "How could such an emotionless voice possibly replace a human?" Little did she expect that today, the voice being cloned would be her own.

She rushed to her workstation without even putting on her shoes, repeatedly confirming that she had not taken any related jobs and tracing the source of the voice.

But in the gray area of AI, protecting her rights feels like fighting with a shadow.

Junjun found a voice identification shop on Taobao through an intermediary, and the identification result showed a similarity of 76%. These types of shops sell over two thousand orders a month. According to the appraiser's assessment, this voice may have blended the tones of multiple voice actors, resembling a deliberately mixed "hybrid voice." This ambiguity left her with no confidence to file a case.

She considered an extreme scenario: if the similarity reached 99%, she would mobilize all her colleagues to flood the brand's comment section to protect her rights. A colleague had previously forced a brand to publicly apologize in this manner.

But this time, it ultimately ended without resolution.

Her colleagues, who had also experienced illegal cloning of their voices, spent six months in court before receiving over thirty thousand yuan in compensation. The highest compensation awarded in the industry had reached 250,000 yuan. Intermediaries also showcase rights protection cases in their social circles, reminding clients to "prohibit sampling and cloning."

Image | Junjun's peers voice their rights in the circle of friends

But in reality, there are too many loopholes.

In the past two years, director John has encountered voiceover contracts that almost always specifically state "sampling prohibited, use for training models prohibited." However, as the sampling required by AI becomes shorter and the barriers lower, as long as an actor has recorded their voice, there is a risk of being "sucked" in, making it difficult to completely guard against.

What is even more disheartening for the industry is the disorder among peers. A female voice actress known to Junjun won over a client during the audition phase with her natural voice, but during the official delivery, she mixed in AI-generated voices and cloned another male voice actor's voice for character dubbing.

After the incident was exposed, the actress insisted that the male voice was "voiced by her husband," causing quite a stir.

According to industry data, the market size of AI voice and semantic technology in China will reach 14.93 billion yuan in 2024. The larger the scale, the more gray area usage there is. In April 2024, national regulatory authorities launched a special action called "Clear and Bright: Rectifying the Abuse of AI Technology," attempting to fill the legal and regulatory gaps.

But before regulations are implemented, many voice actors still feel like they are standing in an exposed wind tunnel.

Dancing with AI

"Voice acting is not just about making sounds; it's about 'being human.'"

In the eyes of national first-class actor Meng Lingjun, the "human touch" supported by life experience is a chasm that AI cannot cross.

Meng Lingjun's voice has left a deep mark on the big screen: from the quirky and lovable Gru in "Despicable Me," to the clever and wise Daniel in "Now You See Me," and to the soul character Alex in "Madagascar 3," which helped the film win the Huabiao Award. In his over twenty-year career, he has infused life into more than 1,500 film and television works.

As a child, Meng Lingjun immersed himself in the joy of imitating film clips and analyzing character tones. In 2001, he officially entered the voiceover industry and began dubbing for theatrical films three years later.

To this day, he still vividly remembers the scene when he dubbed for the movie "Remote Control": the male protagonist in the film is gravely ill and needs to use his last breath to call out to his son.

Before recording, the director specifically reminded him: "Use breath, not real voice."

This meant that every breath had to be carefully controlled, using breath to carry emotion rather than using the throat to exert force. During the process, Meng Lingjun felt as if his chest was being pricked with needles, dizzy and weak, with each breath feeling like it was draining his body. Ultimately, he let out that one "Beng—!" almost exhausting his willpower, leaning against the wall afterward, drenched in cold sweat.

This experience made Meng Lingjun realize that the difficulty of voice acting is never about "how to perform," but whether one can truly empathize. Understanding a character's experiences, personality, psychology, and motivations is what gives voice its life.

For the audience, this "human touch" is equally crucial. Fanfan from Baoding, Hebei, is a seasoned radio drama enthusiast, and she admits that if all works were replaced with AI, she probably wouldn't listen anymore She recalls the days when online voice acting began to emerge in 2008: in forums and message boards, enthusiasts would dub film clips with rudimentary equipment and rough post-production. Every small recording would end in disbandment, and "annual updates" were the norm. The passion of that time drove everyone to pour their thoughts and emotions into their roles and lines.

The turning point came around 2017. Vertical audio platforms like Cat Ear FM and Manbo began to rise, and high-quality commercial audio dramas such as "Sa Ye" and "Mo Du" were successively launched. "Masters" of voice acting and their teams entered the scene strongly, and a paid model began to take shape, costing 20-50 yuan per season, with an entire drama potentially costing one or two hundred or even three hundred yuan. Stable updates, exquisite production, and professional performances made listeners willing to pay.

It was against this industry backdrop that Xuan Xuan's listening experience was completely transformed. When she first got into it in 2019, she spent her entire weekends and after-work hours listening to audio dramas, lying on the window sill with a pillow, never getting tired even after listening to it ten or twenty times. The high-quality voice acting made the characters more vivid, and the interactions brought by offline comic conventions made her feel an emotional connection beyond the voice, a real experience that AI cannot replicate.

The market is still a human market; AI is just a tool. Xuan Xuan once listened to a work authorized by a famous CV to AI, and while the voice was smooth, it lacked genuine emotion, with incorrect emphasis and a thin post-production quality.

Real human voices can always create the "101st" possibility, while AI can only generate results based on existing human samples. Meng Lingjun once declined to sell his voice rights: "They can buy my samples, but they cannot buy my creativity."

AI is not a threat to everyone. In the market where Guangdong voice director John operates, clients require a balance of efficiency, artistry, and budget, simply put, "fast, good, and cheap." AI can easily replace lower-level positions, but for professionals, it remains an auxiliary tool.

"We often train ourselves to be machines, which is why we lose to machines." John believes that the human touch that AI simulates often stems from real instability and imperfection.

The changes in the industry have also opened up survival space for prepared newcomers. Xie Wanwan, during the pandemic, combined AI tools with her team to generate lines in the early stages of animation voice acting, allowing artists to create based on that, and then performed by actors, achieving a balance of efficiency and quality.

Image | Xie Wanwan participating in a voice acting academic seminar

This method not only helped her cope with industry pressure but also made her realize that AI brings not just challenges but sometimes creates new opportunities. Some works with non-standard Mandarin may gain a certain amount of views, but in the end, they can still be easily replaced by better real voices.

For voice actors who know how to leverage their advantages, this means that even in the face of the AI wave, they can still find space to thrive.

Jun Jun, Luo Ying, and Cheng Zi are also exploring ways to survive in the AI wave. Jun Jun maintains stable bargaining power with voice acting in Chinese, Japanese, and English; Luo Ying views AI as a tool, maintaining business volume while continuously improving her skills; Oranges seized the opportunity of short video platforms turning to real human voiceovers, achieving a business rebound.

Industry changes made them realize that AI is not an enemy, but a mirror. It forces voice actors to continuously upgrade their skills and explore the value of human voice.

The choices of the audience ultimately determine the boundaries of technology. FanFan and XuanXuan are still more willing to pay for "people": the interactions at offline comic exhibitions, the emotional resonance of radio dramas, and the life experiences conveyed through voice—these depths cannot be replicated by AI.

Even if AI participates in the early stages of production, the final performance still relies on human actors. Xie Wanwan insists in animation production that while the director may provide reference AI voices, her and the team's work is to interpret with their own emotions.

"The future of the voice acting industry belongs only to those with a soul," Meng Lingjun said.

He believes that amid the industry storm, some are anxious to leave, while others seize the opportunity to grow. The real key is to retain the warmth of the human voice while learning to dance with AI.

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investing based on this is at one's own risk