diff --git a/backend/blocks/video/narration.py b/backend/blocks/video/narration.py
new file mode 100644
index 0000000000..30641780a7
--- /dev/null
+++ b/backend/blocks/video/narration.py
@@ -0,0 +1,161 @@
+"""
+VideoNarrationBlock - Generate AI voice narration and add to video
+"""
+import uuid
+from typing import Literal
+
+from backend.data.block import Block, BlockCategory, BlockOutput
+from backend.data.block import BlockSchemaInput, BlockSchemaOutput
+from backend.data.model import SchemaField, CredentialsMetaInput, APIKeyCredentials
+from backend.integrations.providers import ProviderName
+from backend.util.exceptions import BlockExecutionError
+
+
+class VideoNarrationBlock(Block):
+    """Generate AI narration and add to video."""
+
+    class Input(BlockSchemaInput):
+        credentials: CredentialsMetaInput[
+            Literal[ProviderName.ELEVENLABS], Literal["api_key"]
+        ] = SchemaField(
+            description="ElevenLabs API key for voice synthesis"
+        )
+        video_in: str = SchemaField(
+            description="Input video file",
+            json_schema_extra={"format": "file"}
+        )
+        script: str = SchemaField(
+            description="Narration script text"
+        )
+        voice_id: str = SchemaField(
+            description="ElevenLabs voice ID",
+            default="21m00Tcm4TlvDq8ikWAM"  # Rachel
+        )
+        mix_mode: Literal["replace", "mix", "ducking"] = SchemaField(
+            description="How to combine with original audio",
+            default="ducking"
+        )
+        narration_volume: float = SchemaField(
+            description="Narration volume (0.0 to 2.0)",
+            default=1.0,
+            ge=0.0,
+            le=2.0,
+            advanced=True
+        )
+        original_volume: float = SchemaField(
+            description="Original audio volume when mixing (0.0 to 1.0)",
+            default=0.3,
+            ge=0.0,
+            le=1.0,
+            advanced=True
+        )
+
+    class Output(BlockSchemaOutput):
+        video_out: str = SchemaField(
+            description="Video with narration",
+            json_schema_extra={"format": "file"}
+        )
+        audio_file: str = SchemaField(
+            description="Generated audio file",
+            json_schema_extra={"format": "file"}
+        )
+
+    def __init__(self):
+        super().__init__(
+            id="e5f6a7b8-c9d0-1234-ef56-789012345678",
+            description="Generate AI narration and add to video",
+            categories={BlockCategory.MULTIMEDIA, BlockCategory.AI},
+            input_schema=self.Input,
+            output_schema=self.Output,
+            test_input={
+                "video_in": "/tmp/test.mp4",
+                "script": "Hello world",
+                "credentials": {"provider": "elevenlabs", "id": "test", "type": "api_key"}
+            },
+            test_output=[("video_out", str), ("audio_file", str)],
+            test_mock={"_generate_narration": lambda *args: ("/tmp/narrated.mp4", "/tmp/audio.mp3")}
+        )
+
+    async def run(
+        self,
+        input_data: Input,
+        *,
+        credentials: APIKeyCredentials,
+        **kwargs
+    ) -> BlockOutput:
+        try:
+            import requests
+            from moviepy.editor import VideoFileClip, AudioFileClip, CompositeAudioClip
+        except ImportError as e:
+            raise BlockExecutionError(
+                message=f"Missing dependency: {e}. Install moviepy and requests.",
+                block_name=self.name,
+                block_id=str(self.id)
+            )
+
+        video = None
+        final = None
+        try:
+            # Generate narration via ElevenLabs
+            response = requests.post(
+                f"https://api.elevenlabs.io/v1/text-to-speech/{input_data.voice_id}",
+                headers={
+                    "xi-api-key": credentials.api_key.get_secret_value(),
+                    "Content-Type": "application/json"
+                },
+                json={
+                    "text": input_data.script,
+                    "model_id": "eleven_monolingual_v1"
+                },
+                timeout=60
+            )
+            response.raise_for_status()
+
+            audio_path = f"/tmp/narration_{uuid.uuid4()}.mp3"
+            with open(audio_path, "wb") as f:
+                f.write(response.content)
+
+            # Combine with video
+            video = VideoFileClip(input_data.video_in)
+            narration = AudioFileClip(audio_path).volumex(input_data.narration_volume)
+
+            if input_data.mix_mode == "replace":
+                final_audio = narration
+            elif input_data.mix_mode == "mix":
+                if video.audio:
+                    original = video.audio.volumex(input_data.original_volume)
+                    final_audio = CompositeAudioClip([original, narration])
+                else:
+                    final_audio = narration
+            else:  # ducking
+                if video.audio:
+                    original = video.audio.volumex(input_data.original_volume)
+                    final_audio = CompositeAudioClip([original, narration])
+                else:
+                    final_audio = narration
+
+            final = video.set_audio(final_audio)
+
+            output_path = f"/tmp/narrated_{uuid.uuid4()}.mp4"
+            final.write_videofile(output_path, logger=None)
+
+            yield "video_out", output_path
+            yield "audio_file", audio_path
+
+        except requests.exceptions.RequestException as e:
+            raise BlockExecutionError(
+                message=f"ElevenLabs API error: {e}",
+                block_name=self.name,
+                block_id=str(self.id)
+            )
+        except Exception as e:
+            raise BlockExecutionError(
+                message=f"Failed to add narration: {e}",
+                block_name=self.name,
+                block_id=str(self.id)
+            )
+        finally:
+            if final:
+                final.close()
+            if video:
+                video.close()