@conceptcraft/mindframes 0.3.14 → 0.3.16

package/README.md

@@ -174,7 +174,7 @@ mindframes video generate new my-video --type youtube
 cd my-video
 ```
 
-### 2) Generate assets (voiceover, images, videos, music)
+### 2) Generate early assets (voiceover, images, videos)
 
 ```bash
 mindframes video generate assets
@@ -186,6 +186,9 @@ mindframes video generate assets
 mindframes video generate plan
 ```
 
+After TTS durations and the final timeline are known, generate music/SFX with
+the factual `--request` commands below.
+
 ### 4) Export the video (poster embedded by default)
 
 ```bash
@@ -199,6 +202,21 @@ mindframes video export --no-embed-poster
 mindframes video find "city timelapse"  # search stock videos
 ```
 
+### Video-aware music and SFX
+
+Generate TTS and finalize the timeline first. Then pass factual video data to
+the backend; the CLI does not author provider prompts or music plans.
+
+```bash
+mindframes music generate --request music-request.json --output public/audio/music.mp3
+mindframes sfx generate --request sfx-request.json --output public/audio/logo-lock.mp3
+```
+
+Both JSON files use `schema_version: 1`, `video_description`, `duration_s`, and
+`narration`. An SFX request also includes one `visual_event`; include narration
+word timings when TTS provides them. Gemini turns those final facts into the
+provider-ready music or SFX prompt on the backend.
+
 ### If you prefer using --project
 
 ```bash
@@ -248,7 +266,7 @@ mindframes create "Report" --context-file data.md -o json | jq -r '.viewUrl' | x
 
 ## Requirements
 
-- Node.js 20.0.0 or higher
+- Node.js 22.22.x
 
 ## License