// Word-level speech-to-text transcripts from narration_combined.transcript.json. // Stored as a jsonb array [{word, start, end}] per source to avoid millions of tiny rows. exports.up = (pgm) => { pgm.createTable('transcripts', { id: 'id', project_id: { type: 'integer', notNull: true, references: 'videos', onDelete: 'CASCADE' }, source_name: { type: 'varchar(100)', notNull: true }, // narration_combined | talkinghead words: { type: 'jsonb', notNull: true }, // [{word, start, end}, ...] created_at: { type: 'timestamptz', notNull: true, default: pgm.func('NOW()') }, updated_at: { type: 'timestamptz', notNull: true, default: pgm.func('NOW()') }, }); pgm.addConstraint('transcripts', 'transcripts_project_source_unique', 'UNIQUE (project_id, source_name)'); }; exports.down = (pgm) => { pgm.dropTable('transcripts'); };