@t3lnet/sceneforge 1.0.6 → 1.0.8

package/README.md

@@ -32,8 +32,17 @@ await runDemoFromFile("./examples/create-dxf-quote.yaml", {
 The CLI is included in this package. After install you can run `sceneforge` via `npx` or your package manager:
 
 ```bash
+# Record a demo
 npx sceneforge record --definition examples/create-dxf-quote.yaml --base-url http://localhost:5173
+
+# Run full pipeline
 npx sceneforge pipeline --definition examples/create-dxf-quote.yaml --base-url http://localhost:5173 --clean
+
+# Generate voiceover (with caching enabled by default)
+npx sceneforge voiceover --demo my-demo
+
+# Manage voice cache
+npx sceneforge voice-cache stats
 ```
 
 ## Extension (optional)
@@ -46,9 +55,27 @@ bun run build:extension
 
 Then load the `dist/` folder in `chrome://extensions`.
 
+## Voice Cache
+
+SceneForge automatically caches synthesized voice audio to reduce ElevenLabs API costs. When the same text is synthesized with the same voice and settings, the cached audio is reused instead of making a new API call.
+
+```bash
+# Cache is enabled by default. To disable:
+npx sceneforge voiceover --demo my-demo --no-cache
+
+# Manage the cache:
+npx sceneforge voice-cache stats    # View statistics
+npx sceneforge voice-cache list     # List cached entries
+npx sceneforge voice-cache clear    # Clear the cache
+npx sceneforge voice-cache prune --days 7  # Remove old entries
+```
+
+Cache is stored in `.voice-cache/` in your project root.
+
 ## Notes
 
 - Voiceover generation uses ElevenLabs and requires `ELEVENLABS_API_KEY` + `ELEVENLABS_VOICE_ID`.
+- Voice cache reduces API costs by reusing previously synthesized audio.
 - Video pipeline steps require FFmpeg installed locally.
 
 ## Repository