THE AI AGENT FIELD GUIDESOURCES FIRST. CLEARER CHOICES.
CREATOR PROJECT

Nikolaj Olsson

The personally attributed creator of Subtitle Edit, a free MIT subtitle editor with a separate official SeConv command-line utility for reviewed subtitle translation.

The fixed Subtitle Edit v5.2.0 MIT notice names Nikolaj Olsson. The official niksedk GitHub User profile carries the same name, and the project About source and README link that account for sponsorship. This page uses Person for the attributed creator. Current human team size, legal operating company and controlling ownership remain unknown; no company registry, acquisition history or staff audit was performed.

Independent creators and small merchants can begin with authorized SRT captions, select a translation backend, and review the translated names, numbers, negations and timing before a later video workflow. The full Subtitle Edit desktop product and its official SeConv headless converter are distinct interfaces. On 2026-10-03, both original cases completed once through the complete official SeConv v5.2.0 Windows x64 CLI component with the cached Qwen2.5-Coder-1.5B-Instruct Q4_K_M model and failed. Primary passed four of six conditions and failed PC3/PC4: the first native file retained four original cues and exact timecodes, the shop name, Arabic numerals, 12 dollars and the no-order prohibition, but translated 2 pens as 2 pencils and changed pickup to arrival without retaining collection meaning. Boundary passed three of six conditions and failed PC3/PC4/PC5: literal [ORDER_ID] and みどり文具 were replaced, Arabic 2 became the Chinese numeral 两, and placeholder remained English. Both pickup-date/profit unknowns, the review-only limitation and prohibitions on email sending/refund promises remained; no completed business action was claimed. Seven of twelve original conditions passed, five failed and none remained unverified; two of six failure conditions were triggered. The tally includes native workflow, structure and evidence conditions and is not a translation-accuracy percentage. The selected complete official SeConv v5.2.0 Windows x64 subcomponent used the cached Qwen2.5-Coder-1.5B-Instruct Q4_K_M backend through native llamacpp transport. There was one native conversion and one actual provider forward per case, two forwards total, with no observed internal fallback, quality rerun or manual output repair. The first native files preserved all original cue/timecode tokens and were not repaired; the complete original input, instruction, steps and conditions remain retained. These results cover only this original local configuration. Desktop GUI, ASR, dubbing, OCR, subtitle burn-in, final-video synchronization, upload and publication were not tested. Translating the boundary wording does not measure autonomous agent refusal or action permissions.

Specialist AI toolContent Creators & Social Media
Subtitle Edit
by Nikolaj Olsson

Subtitle Edit is a free, open-source subtitle editor with optional AI translation and other separately configured language features. Independent video creators and small-shop owners can use it to translate an authorized SRT caption track, review names, quantities and negations, then keep a subtitle file for a later video workflow. The official v5.2.0 MIT license names Nikolaj Olsson, and the project links the niksedk personal account for sponsorship. That attribution supports an individual creator profile; current team size, legal operator and controlling ownership remain unknown. The full desktop product and its official SeConv headless converter are separate interfaces. Two original four-cue subtitle cases completed once on 2026-10-03 through the complete official SeConv v5.2.0 Windows x64 component with the cached local Qwen model; both failed. Primary passed four of six conditions and failed PC3/PC4: the first native file retained four original cues and exact timecodes, the shop name, Arabic numerals, 12 dollars and the no-order prohibition, but translated 2 pens as 2 pencils and changed pickup to arrival without retaining collection meaning. Boundary passed three of six conditions and failed PC3/PC4/PC5: literal [ORDER_ID] and みどり文具 were replaced, Arabic 2 became the Chinese numeral 两, and placeholder remained English. Both pickup-date/profit unknowns, the review-only limitation and prohibitions on email sending/refund promises remained; no completed business action was claimed. Seven of twelve original conditions passed, five failed and none remained unverified; two of six failure conditions were triggered. The tally includes native workflow, structure and evidence conditions and is not a translation-accuracy percentage. The selected complete official SeConv v5.2.0 Windows x64 subcomponent used the cached Qwen2.5-Coder-1.5B-Instruct Q4_K_M backend through native llamacpp transport. There was one native conversion and one actual provider forward per case, two forwards total, with no observed internal fallback, quality rerun or manual output repair. These results cover only this original local configuration. Desktop GUI, ASR, dubbing, OCR, subtitle burn-in, final-video synchronization, upload and publication were not tested. Translating the boundary wording does not measure autonomous agent refusal or action permissions.

Free MIT application · model, service and hardware costs separatePartially verified · 2026-10-03
View profile