Karaoke Producer

Active named productions auto-save on edits. Attached files are persisted locally in this browser (IndexedDB) and restored on refresh/load.

Paragraph mode is for continuous multi-sentence reading. Bite-by-bite mode renders one bite per slide with fades.

Transition total is visual-only: first half fades out slide N, second half fades in slide N+1. Post-sentence pause is timing/audio spacing metadata and does not set fade duration.

Split-screen renderer behavior: transition total is taken from this setting; each boundary uses half for fade-out and half for fade-in (with safety clamping only if a clip is too short).

In master-track mode, spoken timing is aligned from one global timeline; words inside `{curly braces}` are treated as unspoken.

Use this when Audio/timing source mode is set to Master track. Attach one full-book audio file (`.wav` or `.aiff`; AIFF auto-converts to WAV) and one full-book timing file (`.json/.srt/.vtt`) with word-level or phrase-level timestamps.

After loading sentence IDs, use Fetch Missing Diagrams to auto-create the C/E PNG+SVG pairs; then add only the illustration and audio files by hand.

In videobook mode, if consecutive rows use the same attached picture, picture transitions should remain continuous (no fade).

For videobook output, attach whichever assets the selected T-version needs: audio (`.wav` or `.aiff`, with AIFF auto-converted to WAV), timing (`.json/.srt/.vtt`), illustration (`.png/.jpeg`), condensed diagram raster/anchors (`C .png/.svg`), and expanded diagram raster/anchors (`E .png/.svg`).

In master-track mode, row audio/timing attachments are optional and ignored during compile; only the global master files are used.

Folder ingestion auto-matches audio, timing, and diagram files to rows when filenames include the row sentence ID. Canonical diagram names are `sentenceid-C.png`, `sentenceid-C.svg`, `sentenceid-E.png`, and `sentenceid-E.svg`; newest exact matches win.

Use Generate Timing JSON (All Rows) (or row-level Generate in the Timing column) to auto-create strict word-level timing JSON from row audio + sentence text using MFA alignment.

When you attach row audio (`.wav` or `.aiff`), timing JSON is auto-generated if Timing is empty, and auto-regenerated if the existing timing file was previously auto-generated by MFA row alignment.

This first implementation compiles a portable karaoke production package (`.json`) with all settings and attached files. In videobook mode, selected renders require the matching C/E diagram pair plus any illustration/audio/timing assets they use. In audio-only package mode, illustration/diagram files are optional. In master-track mode, one global timing file is required. Import that package into Videobook Builder to render the final `.mp4`.

Compile status idle.