Movie Gen Audio
Meta's video-to-audio and text-to-audio generation for cinematic sound.
About
Movie Gen Audio is the sound half of Meta's Movie Gen, a cast of media foundation models announced in October 2024. It generates 48kHz audio, cinematic sound effects and background music, synchronized to a video input or driven by text alone. Meta claims it can produce long-form coherent audio for videos up to several minutes, well beyond the 16-second clips the video model generates.
The paper (arXiv 2410.13720) claims state of the art on video-to-audio and text-to-audio generation, and Meta released Movie Gen Audio Bench, a 527-video evaluation set, so others can test the same tasks. The catch is access: unlike Llama, Meta never open-sourced or productized Movie Gen, and there is no public API or checkpoint. Everything known about quality comes from Meta's own paper and demos.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Movie Gen Audio Bench (527-video eval set) released by Meta; no independent third-party scores published.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: Meta research paper with broad SOTA claims, public evaluation benchmark released. Strengths: 48kHz cinematic audio, video-synced generation, long-form coherence claims. Docked for: (1) no public weights, code or API; (2) all quality claims are provider-reported. License: proprietary research; eval bench is CC-BY-NC, model unreleased.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 5.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 1.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · Announced Oct 2024; paper arXiv 2410.13720
- · 48kHz video-to-audio and text-to-audio; long-form coherence claimed
- · Movie Gen Audio Bench: 527-video eval set (CC-BY-NC)
- · No public weights, code or API; Meta logo staged