An Actor Who Can No Longer Speak, and the AI Voice That Finished His Line
On the evening of October 22, a televised variety gala marking the 90th anniversary of the Long March aired on CCTV. Among its segments was a short dramatic piece titled “Dawn Ahead,” led by one of China’s most respected screen actors, Li Xuejian. According to the show’s chief director, Yu Lei, in an interview that night, Li can no longer hear or speak, and the crew communicated with him by typing on a phone screen. Footage from the broadcast appears to show exactly that: a director holding up a handset, characters appearing, Li reading and nodding.
What made the segment land, and then keep reverberating, was a second disclosure. The narration attributed to Li in the piece was not his voice. It was generated by AI. Yu said no one else could have carried the wordless performance, and that the production team had explored the option at length during preparation.
Li, who is in his early seventies, is a household name in China for roles that span four decades — historical figures, officials, ordinary fathers — and he is one of the country’s most decorated actors, with a shelf of national television and film awards. In recent years his health has been the subject of frequent public concern, and his appearances have grown rare and carefully chosen. That context is why a single line about AI dubbing traveled so fast.
The reaction split along a line that Chinese audiences have been drawing all year. Some viewers found the solution moving: a performer whose body still remembers how to act, lent a voice by software so that a silent man could still be heard. Others found it uncomfortable, arguing that if the audience is moved by a voice the actor did not produce, the tribute is partly a trick — and that the disclosure should have come before the performance, not after.
There is a craft argument underneath. Voice acting is a profession in China, and synthetic speech has already moved into advertising, audiobooks, short dramas and animation, usually framed as a cost saving. When a national broadcaster uses a generated voice for a beloved veteran, the technology stops looking like a bargain and starts looking like a policy question: who gets to decide when a performer’s voice is replaced, and whether the audience is told.
It also lands in a busy season for the industry’s relationship with AI. Chinese studios have been experimenting with AI-assisted dubbing and lip-sync for imported and animated content, and short-drama producers have been open about using synthetic voices to cut production time. Actors’ representatives have begun to ask, quietly, where the boundary sits between assisting a performer and standing in for one.
For now, the segment has done what the gala wanted: people watched, and many cried. The argument that followed is about whether they were crying at a performance or at a reconstruction of one. Li himself, communicating by text, has not publicly weighed in on the choice made on his behalf.
What to watch is whether the disclosure sets a habit. If broadcasters begin labeling synthetic voice work on screen rather than in post-broadcast interviews, the practice becomes a transparent tool. If they don’t, the next controversy will not be about one actor’s health but about how many voices on Chinese television were never in the room.