[richiemcilroy]

Adds a randomized regression test for the "Invalid sequential export audio window" export failure with Studio Sound on.

In 0.6.1, a Studio Sound export of a timeline with a short cut inside one clip failed with that error every time. Turning Studio Sound off made it export. Main already fixes this (594e1b6 and c62021c), but the existing tests only cover a few hand-picked timelines.

Test

studio_sound_random_cuts_never_reject_the_export_window builds random eligible timelines and exports them through the streaming path with Studio Sound on. Each run varies:

• cuts of 0 samples up to about 3 s, with two clips interleaved;
• random mic and system audio offsets of up to ±0.3 s;
• the isolation strength;
• request sizes from 1 sample to 48,001 samples.

It uses voice-like test audio with loud and quiet stretches, so the leveler and denoiser do real work. It fails if any export returns an error or writes a different number of samples than the preview renderer. By default it runs 12 iterations, about 50 s in a debug build. CAP_AUDIO_FUZZ_ITERATIONS and CAP_AUDIO_FUZZ_SEED widen the search locally.

Verification

• Ran 150 iterations on main on an M-series Mac: no window errors.
• Setting STUDIO_SOUND_HISTORY_SAMPLES to 0 recreates the 0.6.1 behaviour (reads only move forward). With that change every iteration fails with "Invalid sequential export audio window", so the test catches the original bug.
• All 25 export_audio tests pass.

Confidence Score: 5/5

The PR appears safe to merge.

Summary

Adds a repeatable randomized test for Studio Sound exports across cuts.

• A randomized test checks Studio Sound exports across different cut patterns.

_{Reviews (1) · Last reviewed commit: "test(editor): fuzz Studio Sound export w..." · Reviewed by Greptile}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
cap-status	Ready	Preview	Oct 26, 2026 9:22pm UTC

[richiemcilroy]

@greptileai review

[capy-ai]

Capy found no issues.