Read the legend 0-1 left to right, and let the mix override segment solo

Read the hovered frame off the legend ramps

Stack the confidence legend to match the two rules under each turn

Rebuild the demo around per-turn confidence

Play through Web Audio so the transport survives Safari

Publish the datasets' own speakers as a Reference pipeline

Link out to the announcement, write-up, benchmarks and product page

Drop the hover bar; the confidence readout is two dots and their values

Drop the mixture playhead line; the progress fill marks the position

Shrink the example cards to a title plus a one-line subheadline

Teal primary colour and a mixture playhead that reads at 1x

Darken the playback fills, most of all on the mixture lane

Reorder examples: LGC first, couch interview first in the real-world row

Keep the mixture playhead visible and label confidence on hover

Cut demo to three 30 s excerpts and add three real-world examples

Remove internal pipeline chain details from UI and data

Deploy Multi-Voice demo

Deploy Multi-Voice demo

Deploy Multi-Voice demo

Deploy Multi-Voice demo

initial commit