이미지: X — 뉴스 앰프
Summary
- Min Choi shared on X a version of the Will Smith spaghetti meme video with sound added using Pika Soundtrack
- Pika Soundtrack is one of four Pika Audio Models released on August 14, adding motion-aware SFX, music, and voice to video
- The other three models are Pika Speech, SFX, and Music, available only through the Pika API Club
- 소개자
- Min Choi(뉴스 앰프), 2026년 8월 14일 X 게시
- 데모 소재
- 윌 스미스 스파게티 먹방 AI 밈 영상에 소리 추가
- 기능
- Pika Soundtrack — 음악·음성·모션 인식 SFX를 영상에 자동 삽입
- 관련 발표
- Pika Audio Models 4종(Soundtrack·Speech·SFX·Music), 8월 14일 API 전용 공개
- 가격 주장
- Pika는 시장 오디오 모델 대비 최대 20배 저렴하다고 밝힘(Soundtrack 초당 0.617)
What came out
A video posted on X by Min Choi, who runs a news amplification account, is drawing attention. The famous AI-generated clip of Will Smith eating spaghetti now has sound added to it. Min Choi described it as "AI fixed it after three years." The tool that added the sound is Pika Soundtrack, a model that automatically layers music, dialogue, and sound effects onto video in sync with on-screen motion. It's known to be one of the audio models Pika recently released.
What this is about
The Will Smith spaghetti video was released in early 2023 and circulated online as a symbol of the limitations of AI video generation at the time. With a distorted face, awkward motion, and no sound at all, it became a widely shared meme representing the "roughness" of early AI video. The version Min Choi shared adds sound to that same meme, effectively serving as a comparison point showing how far AI video technology has come over the past three years.
The tool used to add the sound, Pika Soundtrack, is one of four audio models—Soundtrack, Music, SFX, and Speech—that Pika released on August 14. They handle background music, music generation, sound effects, and voice, respectively, and are reportedly available only through the Pika API Club. Pika claimed at the time that its audio models were up to 20 times cheaper than competing models on the market, and pointed to Hunyuan Foley as a comparable competitor for generating sound from video, saying Soundtrack was twice as cost-efficient. The only specific price disclosed was Soundtrack's rate of $0.617 per second.
Why it matters
AI video generation tools have spent the past few years competing on image quality and motion realism, while sound has largely been an afterthought. The production process, which used to require adding music or sound effects separately to silent video, is being streamlined by the emergence of motion-aware audio models like Pika Soundtrack. Min Choi's comment that "silent AI video is over" reflects this shift. Still, since the model is available only through the API Club, it remains to be seen how accessible it will be for ordinary users to try it out directly.
