docs(video-analyzer): add Video Analyzer application example

site/en/gemini-api/docs/applications/video_analyzer.md

@@ -0,0 +1,88 @@
+# Video Analyzer – Unlock Insights from Your Videos
+
+> **Status:** Draft – initial contribution by @ochoaughini  
+> This page introduces a reference application that demonstrates how to use the Gemini API to transcribe, summarise and search video content. 
+
+## What the app does
+
+* Drag-and-drop a local video (mp4 / mov / webm).  
+* Automatically extracts audio → **speech-to-text** using Gemini audio transcription.  
+* Captures still frames every *N* seconds and sends them to Gemini **multimodal** endpoint for scene description.  
+* Generates:  
+  * An SRT subtitle file.  
+  * A bullet-point **summary** (topics, key moments).  
+  * Embeddings index allowing **semantic search** over the transcript.
+
+## Quick start
+
+```bash
+pip install google-generativeai moviepy ffmpeg-python
+python video_analyzer.py --input my_video.mp4 --model gemini-pro-vision
+```
+
+The script will write:
+
+* `my_video.srt` – subtitles
+* `my_video.summary.txt` – text summary
+* `my_video.index.json` – embedding index for search
+
+## Core code snippets
+
+### 1 · Extract audio and transcribe
+```python
+import google.generativeai as genai
+from moviepy.editor import VideoFileClip
+
+audio_path = "tmp_audio.mp3"
+VideoFileClip(video_path).audio.write_audiofile(audio_path, logger=None)
+
+model = genai.GenerativeModel("gemini-pro")
+transcription = model.generate_content(Path(audio_path).read_bytes(), mime_type="audio/mpeg")
+```
+
+### 2 · Describe video frames
+```python
+from pathlib import Path
+from PIL import Image
+
+def sample_frames(video_path, every_sec=5):
+    clip = VideoFileClip(video_path)
+    for t in range(0, int(clip.duration), every_sec):
+        frame = clip.get_frame(t)
+        img = Image.fromarray(frame)
+        fname = f"frame_{t:04}.png"
+        img.save(fname)
+        yield fname
+
+vision_model = genai.GenerativeModel("gemini-pro-vision")
+
+scene_descriptions = []
+for frame_file in sample_frames(video_path):
+    desc = vision_model.generate_content(Path(frame_file).read_bytes(), mime_type="image/png")
+    scene_descriptions.append(desc.text)
+```
+
+### 3 · Summarise and index
+```python
+summary = model.generate_content(
+    "Summarise this transcript:\n" + transcription.text
+).text
+
+embeddings = model.embed_content(transcription.text.split("\n"))
+```
+
+## Folder layout
+```
+video-analyzer/
+├── video_analyzer.py        # main script
+├── templates/               # optional web UI
+└── README.md                # setup & usage docs
+```
+
+## Next steps
+* Add a Streamlit front-end.  
+* Integrate **Gemini function-calling** for automatic action extraction.  
+* Accept YouTube URLs (download + analyse).
+
+---
+**Contributing** – please feel free to open issues or PRs to improve this example.