Tests we ran ourselves
Tools we ran ourselves on the same short clips, with the timings, the measured result and what failed. Every number comes from our own runs; nothing is copied from a vendor page.
Tests so far
- We tested auto-editor's silence cutting on three dialogue clips: on dialogue with a music score the default setting cut almost nothing; a stricter setting cut into speech.
- We tested faster-whisper captions on three dialogue clips: small.en beat base.en on all three clips against the official subtitles, on a laptop CPU.
- We tested burning captions into video with FFmpeg: all 35 captions landed in the picture in both styles, with the file length unchanged.
- We tested Remotion's TikTok captions template on three clips: it renders a vertical captioned video, but the clip sits in the top 452 rows of a white frame and the captions stay on screen through every gap in speech.
- We tested Remotion's transitions package on three clip pairs: fade, slide and wipe all worked on three clip pairs; fade and the other two place the clip differently in the frame.
- We cut the silence and added captions to three clips, start to finish: the whole job on three clips with auto-editor, faster-whisper and FFmpeg in a row, 92 to 118 seconds per 60-second clip, and what went wrong with the captions.
- One clip, eight caption styles: which one reads best?: the same five seconds of film through seven of our caption styles and FFmpeg's plain burn-in, with contrast, screen coverage and position measured for each.
How we test
- Same inputs every time: three 60-second clips from the open film Tears of Steel (CC-BY 3.0, Blender Foundation), with a checksum for each file so anyone can use the same ones.
- One cold run, then three timed runs per clip and setting. We publish the median, the spread, peak memory and the CPU load on the laptop just before each run.
- A written reference for quality: the film's official English subtitles for words and speech timing; the source frames for anything drawn into the picture.
- Every page has a What failed box. Failures and surprises are part of the result.
- We never publish a time saved, because that would need a manual baseline we have not measured.
What we test next
- WhisperX word timings against faster-whisper on the same clips. On 5 Nov it would not install here: its current releases require Python 3.13 or older (this machine runs 3.14), and the older releases pin a ctranslate2 version with no build for 3.14. Next run: a separate Python 3.12 environment.
- Long-video-to-Shorts tools against highlights we mark by hand first. We looked at both on 6 Nov and could not run either here yet. OpenShorts runs through Docker and its README lists a Google Gemini key as required for its AI features (a local language model can replace it for picking moments only); we run no accounts or keys in these tests. FunClip needs Python 3.12 with PyTorch and its models, and this machine runs Python 3.14. Next run: a Python 3.12 environment for FunClip, and OpenShorts only if it can run with a local model and no key.
- A clean talking-head recording for silence cutting, where auto-editor should do much better than on a film score.
Page checked on Nov 4, 2026.