add vad example

Author: maxkahan

examples/vad_example.py

@@ -0,0 +1,43 @@
+# Stream × Silero — VAD Bot
+
+Detect when people speak in a Stream Video call.
+
+This minimal example starts a bot that listens to the call, runs every audio
+frame through the [Silero VAD](https://github.com/snakers4/silero-vad) model
+and prints each speech turn with its duration.
+
+---
+
+## Quick start
+
+```bash
+cd examples/vad_silero
+
+# install deps (choose one)
+pip install -e .
+uv venv .venv && source .venv/bin/activate && uv sync   # fast ⚡️
+
+cp ../stt_deepgram_transcription/env.example .env  # fill STREAM_* keys
+python main.py                                    # or: uv -m python main.py
+```
+
+Speak in the browser tab – you'll see logs like:
+
+```
+🤖 VAD bot starting – speak in the call and watch the console.
+📞 Call ready: 7a1f…
+[12:03:45] Speech from user-d1e8… — 1.32s
+[12:03:49] Speech from user-d1e8… — 0.87s
+```
+
+---
+
+## How it works
+
+1. Creates two temporary Stream users (human + `vad-bot-*`).
+2. Opens the call URL so you can join immediately.
+3. Every incoming PCM frame goes to `Silero.process_audio()`.
+4. The plugin emits `partial` (in-progress) and `audio` (end-of-turn) events.
+5. On **Ctrl-C** the bot leaves the call and temporary users are deleted.
+
+< 120 lines of Python 🐍 

examples/vad_silero/README.md

@@ -0,0 +1 @@
+ 

examples/vad_silero/__init__.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python3
+"""
+Example: Voice-Activity-Detection bot (Silero VAD)
+
+The script joins a Stream video call with a bot that detects when anyone
+speaks, using the Silero VAD plugin (`getstream.plugins.vad.silero`).
+Each complete speech turn is logged with a timestamp and duration.
+
+Run:
+    python main.py
+
+Environment: copy `examples/env.example` to `.env` and fill in
+`STREAM_API_KEY`, `STREAM_API_SECRET` (and optionally `STREAM_BASE_URL`).
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+import os
+import time
+from typing import Any
+from uuid import uuid4
+
+from dotenv import load_dotenv
+
+from examples.utils import create_user, open_browser
+from getstream.stream import Stream
+from getstream.video import rtc
+from getstream.video.rtc.track_util import PcmData
+from getstream.plugins.vad.silero import Silero
+
+# ---------------------------------------------------------------------------
+# Logging setup – INFO level so we see joins / leaves, etc.
+# ---------------------------------------------------------------------------
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s  %(levelname)s  %(message)s",
+)
+
+
+async def main() -> None:
+    """Create a call and start the Silero VAD bot."""
+
+    # Load env from examples/.env
+    load_dotenv(os.path.join(os.path.dirname(__file__), "..", ".env"))
+
+    client = Stream.from_env()
+
+    human_id = f"user-{uuid4()}"
+    bot_id = f"vad-bot-{uuid4()}"
+
+    create_user(client, human_id, "Human")
+    create_user(client, bot_id, "VAD Bot")
+
+    token = client.create_token(human_id, expiration=3600)
+
+    call_id = str(uuid4())
+    call = client.video.call("default", call_id)
+    call.get_or_create(data={"created_by_id": bot_id})
+
+    logging.info("📞 Call ready: %s", call_id)
+
+    open_browser(client.api_key, token, call_id)
+
+    vad = Silero()
+
+    print("\n🤖 VAD bot starting – speak in the call and watch the console.\n")
+
+    speech_segments: list[dict[str, Any]] = []
+
+    try:
+        async with await rtc.join(call, bot_id) as connection:
+            logging.info("🤖 Bot joined call: %s", call_id)
+
+            # Forward audio frames to the VAD engine
+            @connection.on("audio")
+            async def _on_pcm(pcm: PcmData, user):
+                await vad.process_audio(pcm, user)
+
+            # Complete speech turns
+            @vad.on("audio")  # type: ignore[arg-type]
+            async def _on_turn(pcm: PcmData, user):
+                duration = pcm.duration
+                ts = time.strftime("%H:%M:%S")
+                print(f"[{ts}] Speech from {user} — {duration:.2f}s")
+                speech_segments.append({
+                    "timestamp": ts,
+                    "duration": duration,
+                    "user": user,
+                })
+
+            # Optional: in-progress indicator
+            @vad.on("partial")  # type: ignore[arg-type]
+            async def _on_partial(_: PcmData, user):
+                print(f"    {user} … speaking", end="\r")
+
+            print("🎧 Listening… press Ctrl-C to stop")
+            await connection.wait()
+
+    except (asyncio.CancelledError, KeyboardInterrupt):
+        print("\n⏹️  Stopping VAD bot…")
+    finally:
+        await vad.close()
+
+        print(f"Detected {len(speech_segments)} speech segments")
+        total_duration = sum(segment["duration"] for segment in speech_segments)
+        print(f"Total speech duration: {total_duration:.2f} seconds")
+
+        client.delete_users([human_id, bot_id])
+        print("🧹 Cleanup completed")
+
+
+if __name__ == "__main__":
+    asyncio.run(main()) 

examples/vad_silero/main.py

@@ -0,0 +1,28 @@
+[project]
+name = "vad-silero-example"
+version = "0.1.0"
+description = "Stream Video + Silero VAD demo."
+readme = "README.md"
+requires-python = ">=3.9"
+license = { text = "MIT" }
+
+dependencies = [
+    "python-dotenv>=1.0.0",
+    "torch>=1.11.0",
+    "numpy>=1.20.0",
+    "aiortc>=1.10.1",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=7.0.0",
+    "pytest-asyncio>=0.21.0",
+]
+
+[build-system]
+requires = ["setuptools>=61.0", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[tool.uv.sources]
+getstream = { workspace = true }
+getstream-plugins-vad-silero = { workspace = true }