01
Upload locally
Accept common video and audio formats. The browser creates a local preview immediately, while the production API will create a private R2 upload session.
PRODUCT FLOW
From local video to usable Chinese subtitles.
The product is a focused pipeline: upload a file, transcribe the audio, align pinyin to each Hanzi character, then export subtitles that players can actually use.
02
Transcribe and translate
The processor extracts 16 kHz mono audio, runs Chinese ASR with word timestamps, and adds the selected translation as a separate learning layer.
03
Align subtitles
Pinyin is generated per Hanzi character. Segment timing stays attached to the transcript so users can edit one line without rebuilding the whole file.
04
Export and study
SRT is the compatibility path. Positioned ASS is the high-quality path for players that support per-character placement. Burn-in MP4 can come later.
MVP BOUNDARY
Login and payment are reserved, not blocking the core upload-preview-export loop.
The current front end works as a local-first prototype. Email login, usage limits, Stripe checkout, and cloud jobs can attach to the API contracts already documented in this project.