Allow remotely triggered pipelines for wake word satellites

Add example for ReSpeaker USB 4mic array v2.0

.gitignore

@@ -9,7 +9,7 @@ tmp/
 build
 htmlcov
 
-/.venv/
+.venv/
 .mypy_cache/
 __pycache__/
 

CHANGELOG.md

@@ -1,5 +1,15 @@
 # Changelog
 
+## 1.3.0
+
+- Bump to wyoming 1.5.4 (timers)
+- Add support for voice timers
+- Add `--timer-finished-wav` and `--timer-finished-wav-repeat`
+- Add `--timer-started-command`
+- Add `--timer-updated-command`
+- Add `--timer-cancelled-command`
+- Add `--timer-finished-command`
+
 ## 1.2.0
 
 - Add `--tts-played-command`

README.md

@@ -135,9 +135,12 @@ You can play a WAV file when the wake word is detected (locally or remotely), an
 
 * `--awake-wav <WAV>` - played when the wake word is detected
 * `--done-wav <WAV>` - played when the voice command is finished
+* `--timer-finished-wav <WAV>` - played when a timer is finished
 
 If you want to play audio files other than WAV, use [event commands](#event-commands). Specifically, the `--detection-command` to replace `--awake-wav` and `--transcript-command` to replace `--done-wav`.
 
+The timer finished sound can be repeated with `--timer-finished-wav-repeat <repeats> <delay>` where `<repeats>` is the number of times to repeat the WAV, and `<delay>` is the number of seconds to wait between repeats.
+
 ## Audio Enhancements
 
 Install the dependencies for webrtc:
@@ -179,5 +182,9 @@ Satellites can respond to events from the server by running commands:
 * `--error-command` - an error was sent from the server (text on stdin)
 * `--connected-command` - satellite connected to server
 * `--disconnected-command` - satellite disconnected from server
+* `--timer-started-command` - new timer has started (json on stdin)
+* `--timer-updated-command` - timer has been paused/unpaused or has time added/removed (json on stdin)
+* `--timer-cancelled-command` - timer has been cancelled (timer id on stdin)
+* `--timer-finished-command` - timer has finished (timer id on stdin)
 
 For more advanced scenarios, use an event service (`--event-uri`). See `wyoming_satellite/example_event_client.py` for a basic client that just logs events.

docs/tutorial_2mic.md

@@ -6,7 +6,7 @@ This tutorial should work for almost any Raspberry Pi and USB microphone. Audio
 
 ## Install OS
 
-Follow instructions to [install Raspberry Pi OS](https://www.raspberrypi.com/software/). Under "Choose OS", pick "Raspberry Pi OS (other)" and "Raspberry Pi OS (Legacy, **64-bit**) Lite".
+Follow instructions to [install Raspberry Pi OS](https://www.raspberrypi.com/software/). Under "Choose OS", pick "Raspberry Pi OS (other)" and "Raspberry Pi OS (**64-bit**) Lite".
 
 When asking if you'd like to apply customization settings, choose "Edit Settings" and:
 
@@ -305,6 +305,13 @@ python3 -m venv --system-site-packages .venv
 .venv/bin/pip3 install 'wyoming==1.5.2'
 ```
 
+In case you use a ReSpeaker USB 4mic array v2.0, additionally install pixel-ring:
+
+```sh
+.venv/bin/pip3 install 'pixel-ring'
+```
+
+
 The `--system-site-packages` argument is used to access the pre-installed `gpiozero` and `spidev` Python packages. If these are **not already installed** in your system, run:
 
 ```sh
@@ -385,4 +392,7 @@ Try a voice command and see if the LEDs change. Use `journalctl` to check the lo
 journalctl -u 2mic_leds.service -f
 ```
 
+If you encounter any issues, you can add the `--debug` argument to the command line to increase the log level.
+To control the brightness of the LEDS, use the `--led-brightness ` argument, which accepts integer numbers from 1 to 31.
+
 Make sure to run `sudo systemctl daemon-reload` every time you make changes to the service.

docs/tutorial_installer.md

@@ -4,7 +4,7 @@ Create a voice satellite using a Raspberry Pi 3+ and USB microphone and speakers
 
 ## Install OS
 
-Follow instructions to [install Raspberry Pi OS](https://www.raspberrypi.com/software/). Under "Choose OS", pick "Raspberry Pi OS (other)" and "Raspberry Pi OS (Legacy, **64-bit**) Lite".
+Follow instructions to [install Raspberry Pi OS](https://www.raspberrypi.com/software/). Under "Choose OS", pick "Raspberry Pi OS (other)" and "Raspberry Pi OS (**64-bit**) Lite".
 
 When asking if you'd like to apply customization settings, choose "Edit Settings" and:
 

etc/install-respeaker-drivers.sh

@@ -6,6 +6,14 @@
 
 set -eo pipefail
 
+kernel_formatted="$(uname -r | cut -f1,2 -d.)"
+driver_url_status="$(curl -ILs https://github.com/HinTak/seeed-voicecard/archive/refs/heads/v$kernel_formatted.tar.gz | tac | grep -o "^HTTP.*" | cut -f 2 -d' ' | head -1)"
+
+if  [ ! "$driver_url_status" = 200 ]; then
+echo "Could not find driver for kernel $kernel_formatted"
+exit 1
+fi
+
 apt-get update
 apt-get install --no-install-recommends --yes \
     curl raspberrypi-kernel-headers dkms i2c-tools libasound2-plugins alsa-utils
@@ -22,12 +30,11 @@ pushd "${temp_dir}"
 
 # Download source code to temporary directory
 # NOTE: There are different branches in the repo for different kernel versions.
-# We use v6.1 here.
 echo 'Downloading source code'
-curl -L -o - 'https://github.com/HinTak/seeed-voicecard/archive/refs/heads/v6.1.tar.gz' | \
+curl -L -o - "https://github.com/HinTak/seeed-voicecard/archive/refs/heads/v$kernel_formatted.tar.gz" | \
     tar -xzf -
 
-cd seeed-voicecard-6.1/
+cd seeed-voicecard-"$kernel_formatted"/
 
 # 1. Build kernel module
 echo 'Building kernel module'
@@ -36,12 +43,18 @@ mod='seeed-voicecard'
 src='./'
 kernel="$(uname -r)"
 marker='0.0.0'
+threads="$(getconf _NPROCESSORS_ONLN)"
+memory="$(LANG=C free -m|awk '/^Mem:/{print $2}')"
+
+if  [ "${memory}" -le 512 ] && [ "${threads}" -gt 2 ]; then
+threads=2
+fi
 
 mkdir -p "/usr/src/${mod}-${ver}"
 cp -a "${src}"/* "/usr/src/${mod}-${ver}/"
 
 dkms add -m "${mod}" -v "${ver}"
-dkms build -k "${kernel}" -m "${mod}" -v "${ver}" && {
+dkms build -k "${kernel}" -m "${mod}" -v "${ver}" -j "${threads}" && {
     dkms install --force -k "${kernel}" -m "${mod}" -v "${ver}"
 }
 

examples/2mic_service.py

@@ -23,7 +23,7 @@ from wyoming.server import AsyncEventHandler, AsyncServer
 from wyoming.vad import VoiceStarted
 from wyoming.wake import Detection
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 NUM_LEDS = 3
 LEDS_GPIO = 12
@@ -41,8 +41,14 @@ async def main() -> None:
     """Main entry point."""
     parser = argparse.ArgumentParser()
     parser.add_argument("--uri", required=True, help="unix:// or tcp://")
-    #
     parser.add_argument("--debug", action="store_true", help="Log DEBUG messages")
+    parser.add_argument(
+        "--led-brightness",
+        type=int,
+        choices=range(1, 32),
+        default=31,
+        help="LED brightness (integer from 1 to 31)",
+    )
     args = parser.parse_args()
 
     logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
@@ -54,7 +60,7 @@ async def main() -> None:
     led_power = gpiozero.LED(LEDS_GPIO, active_high=False)
     led_power.on()
 
-    leds = APA102(num_led=NUM_LEDS)
+    leds = APA102(num_led=NUM_LEDS, global_brightness=args.led_brightness)
 
     # Start server
     server = AsyncServer.from_uri(args.uri)
@@ -148,7 +154,7 @@ class APA102:
     def __init__(
         self,
         num_led,
-        global_brightness=MAX_BRIGHTNESS,
+        global_brightness,
         order="rgb",
         bus=0,
         device=1,
@@ -162,6 +168,7 @@ class APA102:
             self.global_brightness = self.MAX_BRIGHTNESS
         else:
             self.global_brightness = global_brightness
+        _LOGGER.debug("LED brightness: %d", self.global_brightness)
 
         self.leds = [self.LED_START, 0, 0, 0] * self.num_led  # Pixel buffer
         self.spi = spidev.SpiDev()  # Init the SPI device

examples/4mic_service.py

@@ -23,7 +23,7 @@ from wyoming.server import AsyncEventHandler, AsyncServer
 from wyoming.vad import VoiceStarted
 from wyoming.wake import Detection
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 NUM_LEDS = 12
 LEDS_GPIO = 5

examples/usbmic_service.py

@@ -0,0 +1,104 @@
+#!/usr/bin/env python3
+"""Controls the LEDs on the ReSpeaker Mic Array v2.0 (USB) ."""
+import argparse
+import asyncio
+import logging
+import time
+from functools import partial
+
+from pixel_ring import pixel_ring
+from wyoming.event import Event
+from wyoming.satellite import (
+    SatelliteConnected,
+    SatelliteDisconnected,
+    StreamingStarted,
+    StreamingStopped,
+)
+from wyoming.server import AsyncEventHandler, AsyncServer
+from wyoming.snd import Played
+from wyoming.vad import VoiceStarted, VoiceStopped
+from wyoming.wake import Detection
+
+_LOGGER = logging.getLogger(__name__)
+
+
+async def main() -> None:
+    """Main entry point."""
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--uri", required=True, help="unix:// or tcp://")
+    parser.add_argument("--debug", action="store_true", help="Log DEBUG messages")
+    args = parser.parse_args()
+
+    logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
+    _LOGGER.debug(args)
+
+    _LOGGER.info("Ready")
+
+    # Turn on power to LEDs
+    pixel_ring.set_color_palette(0x0080FF, 0x007A37)
+    pixel_ring.think()
+    await asyncio.sleep(3)
+    pixel_ring.off()
+
+    # Start server
+    server = AsyncServer.from_uri(args.uri)
+
+    try:
+        await server.run(partial(LEDsEventHandler, args))
+    except KeyboardInterrupt:
+        pass
+    finally:
+        pixel_ring.off()
+
+
+class LEDsEventHandler(AsyncEventHandler):
+    """Event handler for clients."""
+
+    def __init__(
+        self,
+        cli_args: argparse.Namespace,
+        *args,
+        **kwargs,
+    ) -> None:
+        super().__init__(*args, **kwargs)
+
+        self.cli_args = cli_args
+        self.client_id = str(time.monotonic_ns())
+
+        _LOGGER.debug("Client connected: %s", self.client_id)
+
+    async def handle_event(self, event: Event) -> bool:
+        _LOGGER.debug(event)
+
+        if Detection.is_type(event.type):
+            _LOGGER.debug("Detection")
+            pixel_ring.wakeup()
+        elif VoiceStarted.is_type(event.type):
+            _LOGGER.debug("VoiceStarted")
+            pixel_ring.speak()
+        elif VoiceStopped.is_type(event.type):
+            _LOGGER.debug("VoiceStopped")
+            pixel_ring.spin()
+        elif StreamingStopped.is_type(event.type):
+            _LOGGER.debug("StreamingStopped")
+            pixel_ring.off()
+        elif SatelliteConnected.is_type(event.type):
+            _LOGGER.debug("SatelliteConnected")
+            pixel_ring.think()
+            await asyncio.sleep(2)
+            pixel_ring.off()
+        elif Played.is_type(event.type):
+            _LOGGER.debug("Played")
+            pixel_ring.off()
+        elif SatelliteDisconnected.is_type(event.type):
+            _LOGGER.debug("SatelliteDisconnected")
+            pixel_ring.mono(0xFF0000)
+
+        return True
+
+
+if __name__ == "__main__":
+    try:
+        asyncio.run(main())
+    except KeyboardInterrupt:
+        pass

examples/websocket-service/requirements.txt

@@ -0,0 +1 @@
+websockets==12.0

examples/websocket-service/server.py

@@ -0,0 +1,100 @@
+#!/usr/bin/env python3
+import argparse
+import asyncio
+import json
+import logging
+from functools import partial
+from typing import Optional
+
+import websockets
+from wyoming.event import Event
+from wyoming.server import AsyncEventHandler, AsyncServer
+
+_LOGGER = logging.getLogger(__name__)
+
+
+async def main() -> None:
+    """Main entry point."""
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--uri", required=True, help="unix:// or tcp://")
+    parser.add_argument("--websocket-host", default="localhost")
+    parser.add_argument("--websocket-port", type=int, default=8675)
+    #
+    parser.add_argument("--debug", action="store_true", help="Log DEBUG messages")
+    args = parser.parse_args()
+
+    logging.basicConfig(level=logging.DEBUG if args.debug else logging.INFO)
+    _LOGGER.debug(args)
+
+    _LOGGER.info("Ready")
+
+    # Start server
+    server = AsyncServer.from_uri(args.uri)
+    queue: "asyncio.Queue[Optional[Event]]" = asyncio.Queue()
+
+    try:
+        async with websockets.serve(
+            partial(websocket_connected, queue),
+            args.websocket_host,
+            args.websocket_port,
+        ):
+            await server.run(partial(WebsocketEventHandler, args, queue))
+    finally:
+        queue.put_nowait(None)
+
+
+# -----------------------------------------------------------------------------
+
+
+async def websocket_connected(queue: "asyncio.Queue[Optional[Event]]", websocket):
+    try:
+        while True:
+            event = await queue.get()
+            if event is None:
+                # Stop signal
+                break
+
+            await websocket.send(
+                json.dumps(
+                    {"type": event.type, "data": event.data or {}}, ensure_ascii=False
+                )
+            )
+    except websockets.ConnectionClosed:
+        pass
+    except Exception:
+        _LOGGER.exception("Error in websocket handler")
+
+
+# -----------------------------------------------------------------------------
+
+
+class WebsocketEventHandler(AsyncEventHandler):
+    """Event handler for clients."""
+
+    def __init__(
+        self,
+        cli_args: argparse.Namespace,
+        queue: "asyncio.Queue[Optional[Event]]",
+        *args,
+        **kwargs,
+    ) -> None:
+        super().__init__(*args, **kwargs)
+
+        self.cli_args = cli_args
+        self.queue = queue
+
+    async def handle_event(self, event: Event) -> bool:
+        _LOGGER.debug(event)
+        self.queue.put_nowait(event)
+
+        return True
+
+
+# -----------------------------------------------------------------------------
+
+
+if __name__ == "__main__":
+    try:
+        asyncio.run(main())
+    except KeyboardInterrupt:
+        pass

examples/websocket-service/timers.html

@@ -0,0 +1,244 @@
+<!DOCTYPE html>
+<html lang="en">
+
+  <head>
+
+    <meta charset="utf-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
+    <meta name="description" content="Wyoming satellite timer example">
+    <meta name="author" content="Michael Hansen">
+
+    <title>Wyoming Timer Example</title>
+    <style>
+      body {
+          background-color: black;
+      }
+      
+      .timer-container {
+          display: flex;
+          flex-direction: row;
+          justify-content: center;
+          align-items: center;
+      }
+
+      .timer {
+          display: flex;
+          width: 100px;
+          height: auto;
+          flex-wrap:wrap;
+          justify-content: center;
+          align-items: center;
+          background-color: #03a9f4;
+          padding: 15px;
+          border-radius: 10px;
+          box-shadow: 0 0 10px rgba(0, 0, 0, 0.2);
+          color: white;
+          margin-right: 20px
+      }
+
+      .time {
+          font-size: 20px;
+      }
+
+      .message {
+          text-align: center;
+          width: 90px;
+          font-size: 24px;
+          font-weight: bold;
+      }
+
+      .finished {
+          background-color: red;
+      }
+
+      .paused {
+          background-color: yellow;
+          color: black;
+      }
+
+      .blink {
+        animation: blink-animation 1s linear infinite;
+        -webkit-animation: blink-animation 1s linear infinite;
+      }
+      @keyframes blink-animation {
+          0%, 100% {opacity: 1;}
+          50% {opacity: 0;}
+      }
+      @-webkit-keyframes blink-animation {
+          0%, 100% {opacity: 1;}
+          50% {opacity: 0;}
+      }
+    </style>
+  </head>
+
+  <body>
+    <div id="timers" class="timer-container">
+    </div>
+    <audio id="audio-finished" hidden controls loop src="timer_finished.wav"></audio>
+
+    <script>
+      let socket = new WebSocket("ws://localhost:8675")
+      var timers = []
+      var finished_timers = 0
+
+      function q(selector) {return document.querySelector(selector)}
+
+      socket.onopen = function(e) {
+          console.log("Connected")
+      };
+
+      socket.onmessage = function(event) {
+          wyo_event = JSON.parse(event.data)
+          if (wyo_event.type == "timer-started") {
+              let timer_info = wyo_event.data
+              timer_info.is_active = true
+              console.log(timer_info)
+
+              timers.push(timer_info)
+              
+              let timers_elem = q("#timers")
+              let timer_div = document.createElement("div")
+              timer_div.id = "timer-" + timer_info.id
+              timer_div.classList.add("timer")
+
+              let message_div = document.createElement("div")
+              message_div.classList.add("message")
+
+              if (timer_info.name) {
+                  message_div.innerHTML = timer_info.name
+              } else {
+                  if (timer_info.start_hours) {
+                      message_div.innerHTML += timer_info.start_hours + " hr "
+                  }
+                  if (timer_info.start_minutes) {
+                      message_div.innerHTML += timer_info.start_minutes + " min "
+                  }
+                  if (timer_info.start_seconds) {
+                      message_div.innerHTML += timer_info.start_seconds + " sec "
+                  }
+              }
+              timer_div.appendChild(message_div)
+
+              let time_div = document.createElement("div")
+              time_div.classList.add("time")
+
+              let hours = timer_info.start_hours ?? 0
+              let minutes = timer_info.start_minutes ?? 0
+              let seconds = timer_info.start_seconds ?? 0
+
+              time_div.innerHTML =
+                  hours.toString().padStart(2, 0) + ":" +
+                  minutes.toString().padStart(2, 0) + ":" +
+                  seconds.toString().padStart(2, 0);
+
+              timer_div.appendChild(time_div)
+
+              timers_elem.appendChild(timer_div)
+
+              setTimeout(update_timer, 1000, timer_info)
+          }
+          else if (wyo_event.type == "timer-cancelled") {
+              let timer_info = wyo_event.data
+              timers = timers.filter(t => t.id != timer_info.id)
+              let timer_div = q("#timer-" + timer_info.id)
+              if (timer_div) {
+                  timer_div.remove()
+              }
+          }
+          else if (wyo_event.type == "timer-updated") {
+              let timer_info = wyo_event.data
+              for (i = 0; i < timers.length; i++) {
+                  if (timers[i].id == timer_info.id) {
+                      timers[i].is_active = timer_info.is_active
+                      timers[i].total_seconds = timer_info.total_seconds
+                      break
+                  }
+              }
+
+              let timer_div = q("#timer-" + timer_info.id)
+              if (timer_div) {
+                  if (timer_info.is_active) {
+                      timer_div.classList.remove("paused")
+                  } else {
+                      timer_div.classList.add("paused")
+                  }
+              }
+          }
+          else if (wyo_event.type == "timer-finished") {
+              let timer_info = wyo_event.data
+              for (i = 0; i < timers.length; i++) {
+                  if (timers[i].id == timer_info.id) {
+                      timers[i].total_seconds = 0
+                      finished_timers += 1
+                      if (finished_timers == 1) {
+                          q("#audio-finished").play()
+                      }
+                      break
+                  }
+              }
+              timers = timers.filter(t => t.id != timer_info.id)
+              let timer_div = q("#timer-" + timer_info.id)
+              if (timer_div) {
+                  timer_div.classList.add("finished")
+                  timer_div.classList.add("blink")
+
+                  timer_div.onclick = function() {
+                      timer_div.remove()
+                      finished_timers = Math.max(0, finished_timers - 1)
+                      if (finished_timers == 0) {
+                          let audio = q("#audio-finished")
+                          audio.pause()
+                          audio.currentTime = 0
+                      }
+                  };
+              }
+
+              let time_div = q("#timer-" + timer_info.id + " .time")
+              if (time_div) {
+                  time_div.style.opacity = 0
+              }
+          }
+      };
+
+      socket.onclose = function(event) {
+          console.log("Disconnected")
+      };
+
+      socket.onerror = function(error) {
+      };
+
+      function update_timer(timer_info) {
+          if (timer_info.total_seconds < 0) {
+              return
+          }
+
+          let time_div = q("#timer-" + timer_info.id + " .time")
+          if (!time_div) {
+              return
+          }
+          
+          if (!timer_info.is_active) {
+              // Paused
+              setTimeout(update_timer, 1000, timer_info)
+              return
+          }
+          
+          if (timer_info.total_seconds > 0) {
+              timer_info.total_seconds -= 1
+          }
+          let total_minutes = Math.floor(timer_info.total_seconds / 60)
+          
+          let hours = Math.floor(total_minutes / 60)
+          let minutes = total_minutes % 60
+          let seconds = timer_info.total_seconds % 60
+
+          time_div.innerHTML =
+              hours.toString().padStart(2, 0) + ":" +
+              minutes.toString().padStart(2, 0) + ":" +
+              seconds.toString().padStart(2, 0)
+
+          setTimeout(update_timer, 1000, timer_info)
+      };
+    </script>
+  </body>
+</html>

installer/__main__.py

@@ -17,7 +17,7 @@ from .speakers import configure_speakers
 from .wake_word import configure_wake_word
 from .whiptail import ItemType, error, menu, msgbox, passwordbox, run_with_gauge
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 def main() -> None:

installer/const.py

@@ -9,7 +9,7 @@ from typing import List, Optional
 from .dataclasses_json import DataClassJsonMixin
 
 _DIR = Path(__file__).parent
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 PROGRAM_DIR = _DIR.parent
 LOCAL_DIR = PROGRAM_DIR / "local"

installer/microphone.py

@@ -9,7 +9,7 @@ from typing import Dict, List, Optional
 from .const import RECORD_RMS_MIN, RECORD_SECONDS, Settings
 from .whiptail import gauge, inputbox, menu, msgbox, radiolist
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 def configure_microphone(settings: Settings) -> None:

installer/packages.py

@@ -5,7 +5,7 @@ import subprocess
 from .const import PROGRAM_DIR
 from .whiptail import run_with_gauge
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 def packages_installed(*packages) -> bool:

installer/speakers.py

@@ -6,7 +6,7 @@ from typing import List, Optional
 from .const import PROGRAM_DIR, Settings
 from .whiptail import checklist, inputbox, menu, msgbox, radiolist
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 def configure_speakers(settings: Settings) -> None:

installer/whiptail.py

@@ -11,7 +11,7 @@ from .const import HEIGHT, LIST_HEIGHT, TITLE, WIDTH
 
 ItemType = Union[str, Tuple[Any, str]]
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 def whiptail(*args) -> Optional[str]:

requirements.txt

@@ -1,3 +1,3 @@
-wyoming==1.5.3
+wyoming==1.5.4
 zeroconf==0.88.0
 pyring-buffer==1.0.0

script/run_usbmic

@@ -0,0 +1,15 @@
+#!/usr/bin/env python3
+import sys
+import subprocess
+import venv
+from pathlib import Path
+
+_DIR = Path(__file__).parent
+_PROGRAM_DIR = _DIR.parent
+_VENV_DIR = _PROGRAM_DIR / ".venv"
+
+context = venv.EnvBuilder().ensure_directories(_VENV_DIR)
+subprocess.check_call(
+    [context.env_exe, str(_PROGRAM_DIR / "examples" / "usbmic_service.py")]
+    + sys.argv[1:]
+)

tests/test_satellite.py

@@ -30,7 +30,7 @@ from .shared import (
     MicClient,
 )
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 TIMEOUT: Final = 1
 

tests/test_wake_streaming.py

@@ -22,7 +22,7 @@ from wyoming_satellite import (
 
 from .shared import MicClient
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 class WakeClient(AsyncClient):

wyoming_satellite/VERSION

@@ -1 +1 @@
-1.2.0
+1.4.0

wyoming_satellite/__main__.py

@@ -1,12 +1,15 @@
 """Main entry point for Wyoming satellite."""
+
 import argparse
 import asyncio
 import logging
+import shlex
 import sys
 from functools import partial
 from pathlib import Path
 
-from wyoming.info import Attribution, Info, Satellite
+from wyoming.audio import AudioFormat
+from wyoming.info import Attribution, Info, MicProgram, Satellite, SndProgram
 from wyoming.server import AsyncServer, AsyncTcpServer
 
 from . import __version__
@@ -22,9 +25,9 @@ from .settings import (
     MicSettings,
     SatelliteSettings,
     SndSettings,
+    TimerSettings,
     VadSettings,
     WakeSettings,
-    WakeWordAndPipeline,
 )
 from .utils import (
     get_mac_address,
@@ -34,7 +37,7 @@ from .utils import (
     split_command,
 )
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 _DIR = Path(__file__).parent
 
 
@@ -120,9 +123,7 @@ async def main() -> None:
         "--wake-word-name",
         action="append",
         default=[],
-        nargs="+",
+        help="Name of wake word to listen for (requires --wake-uri)",
-        metavar=("name", "pipeline"),
-        help="Name of wake word to listen for and optional pipeline name to run (requires --wake-uri)",
     )
     parser.add_argument("--wake-command", help="Program to run for wake word detection")
     parser.add_argument(
@@ -152,9 +153,14 @@ async def main() -> None:
 
     # Voice activity detector
     parser.add_argument(
-        "--vad", action="store_true", help="Wait for speech before streaming audio"
+        "--vad", action="store_true", help="Enable voice activity detector"
+    )
+    parser.add_argument(
+        "--vad-threshold",
+        type=float,
+        default=0.5,
+        help="Threshold for speech (0-1, default: 0.5)",
     )
-    parser.add_argument("--vad-threshold", type=float, default=0.5)
     parser.add_argument("--vad-trigger-level", type=int, default=1)
     parser.add_argument("--vad-buffer-seconds", type=float, default=2)
     parser.add_argument(
@@ -163,6 +169,7 @@ async def main() -> None:
         default=5.0,
         help="Seconds before going back to waiting for speech when wake word isn't detected",
     )
+    parser.add_argument("--vad-finished-speaking-seconds", type=float, default=0.8)
 
     # External event handlers
     parser.add_argument(
@@ -225,6 +232,23 @@ async def main() -> None:
         "--disconnected-command",
         help="Command to run when disconnected from the server",
     )
+    parser.add_argument(
+        "--timer-started-command",
+        help="Command to run when a timer starts",
+    )
+    parser.add_argument(
+        "--timer-updated-command",
+        help="Command to run when a timer is paused, resumed, or has time added or removed",
+    )
+    parser.add_argument(
+        "--timer-cancelled-command",
+        "--timer-canceled-command",
+        help="Command to run when a timer is cancelled",
+    )
+    parser.add_argument(
+        "--timer-finished-command",
+        help="Command to run when a timer finishes",
+    )
 
     # Sounds
     parser.add_argument(
@@ -233,6 +257,17 @@ async def main() -> None:
     parser.add_argument(
         "--done-wav", help="WAV file to play when voice command is done"
     )
+    parser.add_argument(
+        "--timer-finished-wav", help="WAV file to play when a timer finishes"
+    )
+    parser.add_argument(
+        "--timer-finished-wav-repeat",
+        nargs=2,
+        metavar=("repeat", "delay"),
+        type=float,
+        default=(1, 0),
+        help="Number of times to play timer finished WAV and delay between repeats in seconds",
+    )
 
     # Satellite details
     parser.add_argument("--uri", required=True, help="unix:// or tcp://")
@@ -278,14 +313,14 @@ async def main() -> None:
         try:
             import webrtc_noise_gain  # noqa: F401
         except ImportError:
-            _LOGGER.fatal("Install extras for webrtc")
+            _LOGGER.exception("Extras for webrtc are not installed")
             sys.exit(1)
 
     if needs_silero(args):
         try:
             import pysilero_vad  # noqa: F401
         except ImportError:
-            _LOGGER.fatal("Install extras for silerovad")
+            _LOGGER.exception("Extras for silerovad are not installed")
             sys.exit(1)
 
     if args.awake_wav and (not Path(args.awake_wav).is_file()):
@@ -296,8 +331,9 @@ async def main() -> None:
         _LOGGER.fatal("%s does not exist", args.done_wav)
         sys.exit(1)
 
-    if args.vad and (args.wake_uri or args.wake_command):
+    if args.timer_finished_wav and (not Path(args.timer_finished_wav).is_file()):
-        _LOGGER.warning("VAD is not used with local wake word detection")
+        _LOGGER.fatal("%s does not exist", args.timer_finished_wav)
+        sys.exit(1)
 
     logging.basicConfig(
         level=logging.DEBUG if args.debug else logging.INFO, format=args.log_format
@@ -316,9 +352,46 @@ async def main() -> None:
             attribution=Attribution(name="", url=""),
             installed=True,
             version=__version__,
+            has_vad=args.vad,
+            supports_trigger=True,
+            active_wake_words=args.wake_word_name,
         )
     )
 
+    if args.mic_command:
+        mic_program = shlex.split(args.mic_command)[0]
+        wyoming_info.mic = [
+            MicProgram(
+                name=mic_program,
+                description=args.mic_command,
+                attribution=Attribution(name="", url=""),
+                installed=True,
+                version=None,
+                mic_format=AudioFormat(
+                    rate=args.mic_command_rate,
+                    width=args.mic_command_width,
+                    channels=args.mic_command_channels,
+                ),
+            )
+        ]
+
+    if args.snd_command:
+        snd_program = shlex.split(args.snd_command)[0]
+        wyoming_info.snd = [
+            SndProgram(
+                name=snd_program,
+                description=args.snd_command,
+                attribution=Attribution(name="", url=""),
+                installed=True,
+                version=None,
+                snd_format=AudioFormat(
+                    rate=args.snd_command_rate,
+                    width=args.snd_command_width,
+                    channels=args.snd_command_channels,
+                ),
+            )
+        ]
+
     settings = SatelliteSettings(
         mic=MicSettings(
             uri=args.mic_uri,
@@ -340,16 +413,17 @@ async def main() -> None:
             trigger_level=args.vad_trigger_level,
             buffer_seconds=args.vad_buffer_seconds,
             wake_word_timeout=args.vad_wake_word_timeout,
+            finished_speaking_seconds=args.vad_finished_speaking_seconds,
         ),
         wake=WakeSettings(
             uri=args.wake_uri,
             command=split_command(args.wake_command),
-            names=[
+            names=args.wake_word_name,
-                WakeWordAndPipeline(*wake_name) for wake_name in args.wake_word_name
+            refractory_seconds=(
-            ],
+                args.wake_refractory_seconds
-            refractory_seconds=args.wake_refractory_seconds
+                if args.wake_refractory_seconds > 0
-            if args.wake_refractory_seconds > 0
+                else None
-            else None,
+            ),
         ),
         snd=SndSettings(
             uri=args.snd_uri,
@@ -379,6 +453,15 @@ async def main() -> None:
             connected=split_command(args.connected_command),
             disconnected=split_command(args.disconnected_command),
         ),
+        timer=TimerSettings(
+            started=split_command(args.timer_started_command),
+            updated=split_command(args.timer_updated_command),
+            cancelled=split_command(args.timer_cancelled_command),
+            finished=split_command(args.timer_finished_command),
+            finished_wav=args.timer_finished_wav,
+            finished_wav_plays=int(args.timer_finished_wav_repeat[0]),
+            finished_wav_delay=args.timer_finished_wav_repeat[1],
+        ),
         debug_recording_dir=args.debug_recording_dir,
     )
 

wyoming_satellite/event_handler.py

@@ -9,7 +9,7 @@ from wyoming.server import AsyncEventHandler
 
 from .satellite import SatelliteBase
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 class SatelliteEventHandler(AsyncEventHandler):

wyoming_satellite/example_event_client.py

@@ -8,7 +8,7 @@ from functools import partial
 from wyoming.event import Event
 from wyoming.server import AsyncEventHandler, AsyncServer
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 async def main() -> None:

wyoming_satellite/satellite.py

@@ -1,17 +1,20 @@
 """Satellite code."""
+
 import array
 import asyncio
 import logging
 import math
 import time
 import wave
+from dataclasses import dataclass
 from enum import Enum, auto
 from pathlib import Path
+from queue import Queue
 from typing import Callable, Dict, Final, List, Optional, Set, Union
 
 from pyring_buffer import RingBuffer
 from wyoming.asr import Transcript
-from wyoming.audio import AudioChunk, AudioFormat, AudioStart, AudioStop
+from wyoming.audio import AudioChunk, AudioStart, AudioStop
 from wyoming.client import AsyncClient
 from wyoming.error import Error
 from wyoming.event import Event, async_write_event
@@ -28,6 +31,7 @@ from wyoming.satellite import (
     StreamingStopped,
 )
 from wyoming.snd import Played, SndProcessAsyncClient
+from wyoming.timer import TimerCancelled, TimerFinished, TimerStarted, TimerUpdated
 from wyoming.tts import Synthesize
 from wyoming.vad import VoiceStarted, VoiceStopped
 from wyoming.wake import Detect, Detection, WakeProcessAsyncClient
@@ -40,10 +44,10 @@ from .utils import (
     run_event_command,
     wav_to_events,
 )
-from .vad import SileroVad
+from .vad import SileroVad, VoiceCommandSegmenter
 from .webrtc import WebRtcAudio
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 _PONG_TIMEOUT: Final = 5
 _PING_SEND_DELAY: Final = 2
@@ -59,6 +63,12 @@ class State(Enum):
     STOPPED = auto()
 
 
+@dataclass
+class SoundEvent:
+    event: Event
+    is_tts: bool
+
+
 # -----------------------------------------------------------------------------
 
 
@@ -75,7 +85,7 @@ class SatelliteBase:
         self._mic_task: Optional[asyncio.Task] = None
         self._mic_webrtc: Optional[Callable[[bytes], bytes]] = None
         self._snd_task: Optional[asyncio.Task] = None
-        self._snd_queue: "Optional[asyncio.Queue[Event]]" = None
+        self._snd_queue: "Optional[asyncio.Queue[SoundEvent]]" = None
         self._wake_task: Optional[asyncio.Task] = None
         self._wake_queue: "Optional[asyncio.Queue[Event]]" = None
         self._event_task: Optional[asyncio.Task] = None
@@ -101,6 +111,11 @@ class SatelliteBase:
                 settings.debug_recording_dir, "stt"
             )
 
+        # Names of wake words to listen for
+        self._wake_names: Optional[List[str]] = None
+        if self.settings.wake.names:
+            self._wake_names = self.settings.wake.names
+
     @property
     def is_running(self) -> bool:
         """True if not stopping/stopped."""
@@ -242,6 +257,8 @@ class SatelliteBase:
 
     async def event_from_server(self, event: Event) -> None:
         """Called when an event is received from the server."""
+        forward_event = True
+
         if Ping.is_type(event.type):
             # Respond with pong
             ping = Ping.from_event(event)
@@ -251,12 +268,16 @@ class SatelliteBase:
                 # Enable pinging
                 self._enable_ping()
                 _LOGGER.debug("Ping enabled")
+
+            forward_event = False
         elif Pong.is_type(event.type):
             # Response from our ping
             self._pong_received_event.set()
+            forward_event = False
         elif AudioChunk.is_type(event.type):
             # TTS audio
             await self.event_to_snd(event)
+            forward_event = False
         elif AudioStart.is_type(event.type):
             # TTS started
             await self.event_to_snd(event)
@@ -265,13 +286,6 @@ class SatelliteBase:
             # TTS stopped
             await self.event_to_snd(event)
             await self.trigger_tts_stop()
-        elif Detect.is_type(event.type):
-            # Wake word detection started
-            await self.trigger_detect()
-        elif Detection.is_type(event.type):
-            # Wake word detected
-            _LOGGER.debug("Wake word detected")
-            await self.trigger_detection(Detection.from_event(event))
         elif VoiceStarted.is_type(event.type):
             # STT start
             await self.trigger_stt_start()
@@ -289,39 +303,55 @@ class SatelliteBase:
         elif Error.is_type(event.type):
             _LOGGER.warning(event)
             await self.trigger_error(Error.from_event(event))
+        elif TimerStarted.is_type(event.type):
+            _LOGGER.debug(event)
+            await self.trigger_timer_started(TimerStarted.from_event(event))
+        elif TimerUpdated.is_type(event.type):
+            _LOGGER.debug(event)
+            await self.trigger_timer_updated(TimerUpdated.from_event(event))
+        elif TimerCancelled.is_type(event.type):
+            _LOGGER.debug(event)
+            await self.trigger_timer_cancelled(TimerCancelled.from_event(event))
+        elif TimerFinished.is_type(event.type):
+            _LOGGER.debug(event)
+            await self.trigger_timer_finished(TimerFinished.from_event(event))
 
-        # Forward everything except audio to event service
+        # Forward everything except audio/ping/pong to event service
-        if not AudioChunk.is_type(event.type):
+        if forward_event:
             await self.forward_event(event)
 
-    async def _send_run_pipeline(self, pipeline_name: Optional[str] = None) -> None:
+    async def _send_run_pipeline(
+        self,
+        start_stage: Optional[PipelineStage] = None,
+        end_stage: Optional[PipelineStage] = None,
+        wake_word_name: Optional[str] = None,
+    ) -> None:
         """Sends a RunPipeline event with the correct stages."""
-        if self.settings.wake.enabled:
+        restart_on_end = False
-            # Local wake word detection
-            start_stage = PipelineStage.ASR
-            restart_on_end = False
-        else:
-            # Remote wake word detection
-            start_stage = PipelineStage.WAKE
-            restart_on_end = not self.settings.vad.enabled
 
-        if self.settings.snd.enabled:
+        if start_stage is None:
-            # Play TTS response
+            if self.settings.wake.enabled:
-            end_stage = PipelineStage.TTS
+                # Local wake word detection
-        else:
+                start_stage = PipelineStage.ASR
-            # No audio output
+                restart_on_end = False
-            end_stage = PipelineStage.HANDLE
+            else:
+                # Remote wake word detection
+                start_stage = PipelineStage.WAKE
+                restart_on_end = not self.settings.vad.enabled
+
+        if end_stage is None:
+            if self.settings.snd.enabled:
+                # Play TTS response
+                end_stage = PipelineStage.TTS
+            else:
+                # No audio output
+                end_stage = PipelineStage.HANDLE
 
         run_pipeline = RunPipeline(
             start_stage=start_stage,
             end_stage=end_stage,
-            name=pipeline_name,
+            wake_word_name=wake_word_name,
             restart_on_end=restart_on_end,
-            snd_format=AudioFormat(
-                rate=self.settings.snd.rate,
-                width=self.settings.snd.width,
-                channels=self.settings.snd.channels,
-            ),
         ).event()
         _LOGGER.debug(run_pipeline)
         await self.event_to_server(run_pipeline)
@@ -450,7 +480,10 @@ class SatelliteBase:
                     mic_client = self._make_mic_client()
                     assert mic_client is not None
                     await mic_client.connect()
-                    _LOGGER.debug("Connected to mic service")
+                    _LOGGER.info(
+                        "Connected to mic service: %s",
+                        self.settings.mic.uri or self.settings.mic.command,
+                    )
 
                 event = await mic_client.read_event()
                 if event is None:
@@ -488,9 +521,11 @@ class SatelliteBase:
                     event = AudioChunk(
                         rate=chunk.rate,
                         width=chunk.width,
-                        channels=chunk.channels
+                        channels=(
-                        if (self.settings.mic.channel_index is None)
+                            chunk.channels
-                        else 1,
+                            if (self.settings.mic.channel_index is None)
+                            else 1
+                        ),
                         audio=audio_bytes,
                     ).event()
                 else:
@@ -524,10 +559,10 @@ class SatelliteBase:
     # Sound
     # -------------------------------------------------------------------------
 
-    async def event_to_snd(self, event: Event) -> None:
+    async def event_to_snd(self, event: Event, is_tts: bool = True) -> None:
         """Send an event to the sound service."""
         if self._snd_queue is not None:
-            self._snd_queue.put_nowait(event)
+            self._snd_queue.put_nowait(SoundEvent(event, is_tts))
 
     def _make_snd_client(self) -> Optional[AsyncClient]:
         """Create client for snd service."""
@@ -562,13 +597,17 @@ class SatelliteBase:
                 if self._snd_queue is None:
                     self._snd_queue = asyncio.Queue()
 
-                event = await self._snd_queue.get()
+                snd_event = await self._snd_queue.get()
+                event = snd_event.event
 
                 if snd_client is None:
                     snd_client = self._make_snd_client()
                     assert snd_client is not None
                     await snd_client.connect()
-                    _LOGGER.debug("Connected to snd service")
+                    _LOGGER.info(
+                        "Connected to snd service: %s",
+                        self.settings.snd.uri or self.settings.snd.command,
+                    )
 
                 # Audio processing
                 if self.settings.snd.needs_processing and AudioChunk.is_type(
@@ -589,7 +628,10 @@ class SatelliteBase:
                     event.type
                 ):
                     await _disconnect()
-                    await self.trigger_played()
+                    if snd_event.is_tts:
+                        await self.trigger_played()
+                        await self.event_to_server(Played().event())
+
                     snd_client = None  # reconnect on next event
             except asyncio.CancelledError:
                 break
@@ -633,9 +675,8 @@ class SatelliteBase:
             for event in wav_to_events(
                 wav_path,
                 samples_per_chunk=self.settings.snd.samples_per_chunk,
-                volume_multiplier=self.settings.snd.volume_multiplier,
             ):
-                await self.event_to_snd(event)
+                await self.event_to_snd(event, is_tts=False)
         except Exception:
             # Unmute in case of an error
             self.microphone_muted = False
@@ -709,7 +750,10 @@ class SatelliteBase:
                     wake_client = self._make_wake_client()
                     assert wake_client is not None
                     await wake_client.connect()
-                    _LOGGER.debug("Connected to wake service")
+                    _LOGGER.info(
+                        "Connected to wake service: %s",
+                        self.settings.wake.uri or self.settings.wake.command,
+                    )
 
                     # Reset
                     from_client_task = None
@@ -773,11 +817,7 @@ class SatelliteBase:
 
     async def _send_wake_detect(self) -> None:
         """Inform wake word service of which wake words to detect."""
-        wake_names: Optional[List[str]] = None
+        await self.event_to_wake(Detect(names=self._wake_names).event())
-        if self.settings.wake.names:
-            wake_names = [w.name for w in self.settings.wake.names]
-
-        await self.event_to_wake(Detect(names=wake_names).event())
         await self.trigger_detect()
 
     # -------------------------------------------------------------------------
@@ -852,6 +892,28 @@ class SatelliteBase:
         """Called when an error occurs on the server."""
         await run_event_command(self.settings.event.error, error.text)
 
+    async def trigger_timer_started(self, timer_started: TimerStarted) -> None:
+        """Called when timer-started event is received."""
+        await run_event_command(self.settings.timer.started, timer_started)
+
+    async def trigger_timer_updated(self, timer_updated: TimerUpdated) -> None:
+        """Called when timer-updated event is received."""
+        await run_event_command(self.settings.timer.updated, timer_updated)
+
+    async def trigger_timer_cancelled(self, timer_cancelled: TimerCancelled) -> None:
+        """Called when timer-cancelled event is received."""
+        await run_event_command(self.settings.timer.cancelled, timer_cancelled.id)
+
+    async def trigger_timer_finished(self, timer_finished: TimerFinished) -> None:
+        """Called when timer-finished event is received."""
+        await run_event_command(self.settings.timer.finished, timer_finished.id)
+        for _ in range(self.settings.timer.finished_wav_plays):
+            await self._play_wav(
+                self.settings.timer.finished_wav,
+                mute_microphone=self.settings.mic.mute_during_awake_wav,
+            )
+            await asyncio.sleep(self.settings.timer.finished_wav_delay)
+
     async def forward_event(self, event: Event) -> None:
         """Forward an event to the event service."""
         if self._event_queue is not None:
@@ -886,7 +948,10 @@ class SatelliteBase:
                     event_client = self._make_event_client()
                     assert event_client is not None
                     await event_client.connect()
-                    _LOGGER.debug("Connected to event service")
+                    _LOGGER.info(
+                        "Connected to event service: %s",
+                        self.settings.event.uri or self.settings.event.command,
+                    )
 
                 await event_client.write_event(event)
             except asyncio.CancelledError:
@@ -935,10 +1000,17 @@ class AlwaysStreamingSatellite(SatelliteBase):
         elif PauseSatellite.is_type(event.type):
             self.is_streaming = False
             _LOGGER.info("Satellite paused")
+        elif Detect.is_type(event.type):
+            # Wake word detection started
+            await self.trigger_detect()
         elif Detection.is_type(event.type):
             # Start debug recording
             if self.stt_audio_writer is not None:
                 self.stt_audio_writer.start()
+
+            # Wake word detected
+            _LOGGER.debug("Wake word detected")
+            await self.trigger_detection(Detection.from_event(event))
         elif Transcript.is_type(event.type) or Error.is_type(event.type):
             # Stop debug recording
             if self.stt_audio_writer is not None:
@@ -1010,10 +1082,16 @@ class VadStreamingSatellite(SatelliteBase):
         if RunSatellite.is_type(event.type):
             self._is_paused = False
             _LOGGER.info("Waiting for speech")
+        elif Detect.is_type(event.type):
+            # Wake word detection started
+            await self.trigger_detect()
         elif Detection.is_type(event.type):
             # Start debug recording
             if self.stt_audio_writer is not None:
                 self.stt_audio_writer.start()
+
+            _LOGGER.debug("Wake word detected")
+            await self.trigger_detection(Detection.from_event(event))
         elif (
             Transcript.is_type(event.type)
             or Error.is_type(event.type)
@@ -1150,8 +1228,17 @@ class WakeStreamingSatellite(SatelliteBase):
         # wake word id -> seconds
         self.refractory_timestamp: Dict[Optional[str], float] = {}
 
+        # Local VAD for detecting the end of a voice command.
+        # If this is disabled, the server is expected to do it remotely.
+        self.vad: Optional[SileroVad] = None
         if settings.vad.enabled:
-            _LOGGER.warning("VAD is enabled but will not be used")
+            _LOGGER.debug("Loading VAD model")
+            self.vad = SileroVad(threshold=settings.vad.threshold, trigger_level=1)
+            _LOGGER.info("Loaded VAD model")
+
+        self.segmenter = VoiceCommandSegmenter(
+            silence_seconds=settings.vad.finished_speaking_seconds
+        )
 
         # Used for debug audio recording so both wake and stt WAV files have the
         # same timestamp.
@@ -1162,6 +1249,9 @@ class WakeStreamingSatellite(SatelliteBase):
         self._wake_info: Optional[Info] = None
         self._wake_info_ready = asyncio.Event()
 
+        self._run_pipeline_queue: Queue[RunPipeline] = Queue()
+        self._next_run_pipeline: Optional[RunPipeline] = None
+
     async def event_from_server(self, event: Event) -> None:
         # Only check event types once
         is_run_satellite = False
@@ -1172,13 +1262,42 @@ class WakeStreamingSatellite(SatelliteBase):
         if RunSatellite.is_type(event.type):
             is_run_satellite = True
             self._is_paused = False
-
         elif PauseSatellite.is_type(event.type):
             is_pause_satellite = True
+        elif Detect.is_type(event.type):
+            # Request to change active wake words
+            detect = Detect.from_event(event)
+            self._wake_names = detect.names
+
+            if not self.is_streaming:
+                await self._send_wake_detect()
+
+            _LOGGER.debug("Updated wake word names: %s", self._wake_names)
         elif Transcript.is_type(event.type):
             is_transcript = True
+            if (self._next_run_pipeline is not None) and (
+                self._next_run_pipeline.end_stage == PipelineStage.ASR
+            ):
+                # Done with remotely triggered pipeline
+                self._next_run_pipeline = None
         elif Error.is_type(event.type):
             is_error = True
+        elif RunPipeline.is_type(event.type):
+            run_pipeline = RunPipeline.from_event(event)
+            if not self.is_streaming:
+                # Run right now
+                self._next_run_pipeline = run_pipeline
+                await self._trigger_pipeline(run_pipeline)
+                return
+
+            # Run after streaming is finished
+            self._run_pipeline_queue.put(run_pipeline)
+        elif AudioStop.is_type(event.type):
+            if (self._next_run_pipeline is not None) and (
+                self._next_run_pipeline.end_stage == PipelineStage.TTS
+            ):
+                # Done with remotely triggered pipeline
+                self._next_run_pipeline = None
 
         if is_transcript or is_pause_satellite:
             # Stop streaming before event_from_server is called because it will
@@ -1199,20 +1318,20 @@ class WakeStreamingSatellite(SatelliteBase):
                 self._is_paused = True
                 _LOGGER.debug("Satellite is paused")
             else:
-                # Go back to wake word detection
+                # Go back to wake word detection or run next pipeline
                 await self.trigger_streaming_stop()
 
                 # It's possible to be paused in the middle of streaming
-                if not self._is_paused:
+                if self._is_paused:
-                    await self._send_wake_detect()
+                    return
-                    _LOGGER.info("Waiting for wake word")
 
-                    # Start debug recording (wake)
+                if self._run_pipeline_queue.empty():
-                    self._debug_recording_timestamp = time.monotonic_ns()
+                    # Back to wake word detection
-                    if self.wake_audio_writer is not None:
+                    await self._start_wake_word_detection()
-                        self.wake_audio_writer.start(
+                else:
-                            timestamp=self._debug_recording_timestamp
+                    # Next run from queue
-                        )
+                    self._next_run_pipeline = self._run_pipeline_queue.get()
+                    await self._trigger_pipeline(self._next_run_pipeline)
 
     async def trigger_server_disonnected(self) -> None:
         await super().trigger_server_disonnected()
@@ -1248,13 +1367,26 @@ class WakeStreamingSatellite(SatelliteBase):
                 self.stt_audio_writer.write(audio_bytes)
 
         if self.is_streaming:
+            if self.vad is not None:
+                if audio_bytes is None:
+                    chunk = AudioChunk.from_event(event)
+                    audio_bytes = chunk.audio
+
+                is_speech = self.vad(audio_bytes)
+                chunk_seconds = (len(audio_bytes) // 2) / 16000
+                if not self.segmenter.process(chunk_seconds, is_speech):
+                    # Send empty audio chunk to indicate command is done
+                    event = AudioChunk(
+                        rate=16000, width=2, channels=1, audio=bytes()
+                    ).event()
+
             # Forward to server
             await self.event_to_server(event)
         else:
             # Forward to wake word service
             await self.event_to_wake(event)
 
-    async def event_from_wake(self, event: Event) -> None:
+    async def event_from_wake(self, event: Event, is_fake: bool = False) -> None:
         if Info.is_type(event.type):
             self._wake_info = Info.from_event(event)
             self._wake_info_ready.set()
@@ -1264,53 +1396,85 @@ class WakeStreamingSatellite(SatelliteBase):
             # Not detecting or no server connected
             return
 
-        if Detection.is_type(event.type):
+        if not Detection.is_type(event.type):
-            detection = Detection.from_event(event)
+            return
 
-            # Check refractory period to avoid multiple back-to-back detections
+        detection = Detection.from_event(event)
-            refractory_timestamp = self.refractory_timestamp.get(detection.name)
+
-            if (refractory_timestamp is not None) and (
+        if (self._next_run_pipeline is not None) and (
-                refractory_timestamp > time.monotonic()
+            self._next_run_pipeline.end_stage == PipelineStage.WAKE
+        ):
+            # Remote pipeline was triggered with only wake word detection
+            wake_word_name: Optional[str] = None
+            if detection.name:
+                wake_word_name = normalize_wake_word(detection.name)
+
+            if (not self._next_run_pipeline.wake_word_names) or (
+                wake_word_name in self._next_run_pipeline.wake_word_names
             ):
-                _LOGGER.debug("Wake word detection occurred during refractory period")
+                self._next_run_pipeline = None
-                return
+                await self.event_to_server(event)
-
+                await self._send_wake_detect()
-            # Stop debug recording (wake)
+                _LOGGER.info("Waiting for wake word")
-            if self.wake_audio_writer is not None:
-                self.wake_audio_writer.stop()
-
-            # Start debug recording (stt)
-            if self.stt_audio_writer is not None:
-                self.stt_audio_writer.start(timestamp=self._debug_recording_timestamp)
-
-            _LOGGER.debug(detection)
-
-            self.is_streaming = True
-            _LOGGER.debug("Streaming audio")
-
-            if self.settings.wake.refractory_seconds is not None:
-                # Another detection may not occur for this wake word until
-                # refractory period is over.
-                self.refractory_timestamp[detection.name] = (
-                    time.monotonic() + self.settings.wake.refractory_seconds
-                )
             else:
-                # No refractory period
+                _LOGGER.debug(
-                self.refractory_timestamp.pop(detection.name, None)
+                    "Got %s but listening for %s",
+                    wake_word_name,
+                    self._next_run_pipeline.wake_word_names,
+                )
+            return
 
+        # Check refractory period to avoid multiple back-to-back detections
+        refractory_timestamp = self.refractory_timestamp.get(detection.name)
+        if (refractory_timestamp is not None) and (
+            refractory_timestamp > time.monotonic()
+        ):
+            _LOGGER.debug("Wake word detection occurred during refractory period")
+            return
+
+        # Stop debug recording (wake)
+        if self.wake_audio_writer is not None:
+            self.wake_audio_writer.stop()
+
+        # Start debug recording (stt)
+        if self.stt_audio_writer is not None:
+            self.stt_audio_writer.start(timestamp=self._debug_recording_timestamp)
+
+        _LOGGER.debug(detection)
+
+        self.segmenter.reset()
+        self.is_streaming = True
+        _LOGGER.debug("Streaming audio")
+
+        if (not is_fake) and (self.settings.wake.refractory_seconds is not None):
+            # Another detection may not occur for this wake word until
+            # refractory period is over.
+            self.refractory_timestamp[detection.name] = (
+                time.monotonic() + self.settings.wake.refractory_seconds
+            )
+        else:
+            # No refractory period
+            self.refractory_timestamp.pop(detection.name, None)
+
+        if not is_fake:
             # Forward to the server
             await self.event_to_server(event)
 
-            # Match detected wake word name with pipeline name
+        start_stage: Optional[PipelineStage] = None
-            pipeline_name: Optional[str] = None
+        end_stage: Optional[PipelineStage] = None
-            if self.settings.wake.names:
-                detection_name = normalize_wake_word(detection.name)
-                for wake_name in self.settings.wake.names:
-                    if normalize_wake_word(wake_name.name) == detection_name:
-                        pipeline_name = wake_name.pipeline
-                        break
 
-            await self._send_run_pipeline(pipeline_name=pipeline_name)
+        if self._next_run_pipeline is not None:
+            # Remote pipeline has been triggered
+            start_stage = self._next_run_pipeline.start_stage
+            end_stage = self._next_run_pipeline.end_stage
+
+        await self._send_run_pipeline(
+            start_stage=start_stage,
+            end_stage=end_stage,
+            wake_word_name=detection.name,
+        )
+
+        if not is_fake:
             await self.forward_event(event)  # forward to event service
             await self.trigger_detection(Detection.from_event(event))
             await self.trigger_streaming_start()
@@ -1330,3 +1494,34 @@ class WakeStreamingSatellite(SatelliteBase):
                 info.wake = self._wake_info.wake
         except asyncio.TimeoutError:
             _LOGGER.warning("Failed to get info from wake service")
+
+    async def _start_wake_word_detection(self) -> None:
+        await self._send_wake_detect()
+        _LOGGER.info("Waiting for wake word")
+
+        # Start debug recording (wake)
+        self._debug_recording_timestamp = time.monotonic_ns()
+        if self.wake_audio_writer is not None:
+            self.wake_audio_writer.start(timestamp=self._debug_recording_timestamp)
+
+    async def _trigger_pipeline(self, run_pipeline: RunPipeline) -> None:
+        _LOGGER.debug("Pipeline remotely triggered: %s", run_pipeline)
+
+        if run_pipeline.start_stage == PipelineStage.WAKE:
+            if run_pipeline.wake_word_names:
+                run_pipeline.wake_word_names = [
+                    normalize_wake_word(name) for name in run_pipeline.wake_word_names
+                ]
+
+            _LOGGER.debug(
+                "Listening once for wake words: %s", run_pipeline.wake_word_names
+            )
+            await self.event_to_wake(Detect(names=run_pipeline.wake_word_names).event())
+        elif run_pipeline.start_stage == PipelineStage.ASR:
+            _LOGGER.debug("Listening for voice command once")
+
+            # Fake a detection
+            await self.event_from_wake(Detection().event(), is_fake=True)
+        elif run_pipeline.start_stage == PipelineStage.TTS:
+            _LOGGER.debug("Speaking once: %s", run_pipeline.announce_text)
+            await self.event_to_server(run_pipeline.event())

wyoming_satellite/settings.py

@@ -1,4 +1,5 @@
 """Satellite settings."""
+
 from abc import ABC
 from dataclasses import dataclass, field
 from pathlib import Path
@@ -103,19 +104,11 @@ class SndSettings(ServiceSettings):
         return self.enabled and (self.volume_multiplier != 1.0)
 
 
-@dataclass(frozen=True)
-class WakeWordAndPipeline:
-    """Wake word name + optional pipeline name."""
-
-    name: str
-    pipeline: Optional[str] = None
-
-
 @dataclass(frozen=True)
 class WakeSettings(ServiceSettings):
     """Wake word service settings."""
 
-    names: Optional[List[WakeWordAndPipeline]] = None
+    names: Optional[List[str]] = None
     """List of wake word names to listen for."""
 
     rate: int = 16000
@@ -150,6 +143,9 @@ class VadSettings:
     wake_word_timeout: Optional[float] = 5.0
     """Seconds before going back to sleep if wake word is not detected."""
 
+    finished_speaking_seconds: float = 0.8
+    """Seconds to wait after speech before voice command is finished."""
+
 
 @dataclass(frozen=True)
 class EventSettings(ServiceSettings):
@@ -172,6 +168,32 @@ class EventSettings(ServiceSettings):
     disconnected: Optional[List[str]] = None
 
 
+@dataclass(frozen=True)
+class TimerSettings:
+    """Voice timer settings."""
+
+    started: Optional[List[str]] = None
+    """Command to run when a timer starts."""
+
+    updated: Optional[List[str]] = None
+    """Command to run when a timer is paused, resumed, or has time added or removed."""
+
+    cancelled: Optional[List[str]] = None
+    """Command to run when a timer is cancelled."""
+
+    finished: Optional[List[str]] = None
+    """Command to run when a timer finishes."""
+
+    finished_wav: Optional[str] = None
+    """WAV file to play when a timer finishes."""
+
+    finished_wav_plays: int = 1
+    """Number of times to play finished WAV."""
+
+    finished_wav_delay: float = 0
+    """Delay in seconds between repeats of finished WAV."""
+
+
 @dataclass(frozen=True)
 class SatelliteSettings:
     """Wyoming satellite settings."""
@@ -181,6 +203,7 @@ class SatelliteSettings:
     wake: WakeSettings = field(default_factory=WakeSettings)
     snd: SndSettings = field(default_factory=SndSettings)
     event: EventSettings = field(default_factory=EventSettings)
+    timer: TimerSettings = field(default_factory=TimerSettings)
 
     restart_timeout: float = 5.0
 

wyoming_satellite/utils/audio.py

@@ -10,7 +10,7 @@ from pyring_buffer import RingBuffer
 from wyoming.audio import AudioChunk, AudioStart, AudioStop
 from wyoming.event import Event
 
-_LOGGER = logging.getLogger()
+_LOGGER = logging.getLogger(__name__)
 
 
 class AudioBuffer:

wyoming_satellite/utils/misc.py

@@ -1,24 +1,32 @@
 """Miscellaneous utilities."""
 import argparse
 import asyncio
+import json
 import logging
 import re
 import shlex
 import unicodedata
 import uuid
 from functools import lru_cache
-from typing import List, Optional
+from typing import List, Optional, Union
 
-_LOGGER = logging.getLogger()
+from wyoming.event import Eventable
+
+_LOGGER = logging.getLogger(__name__)
 
 
 async def run_event_command(
-    command: Optional[List[str]], command_input: Optional[str] = None
+    command: Optional[List[str]], command_input: Optional[Union[str, Eventable]] = None
 ) -> None:
     """Run a custom event command with optional input."""
     if not command:
         return
 
+    if isinstance(command_input, Eventable):
+        # Convert event to JSON
+        event_dict = command_input.event().to_dict()
+        command_input = json.dumps(event_dict, ensure_ascii=False)
+
     _LOGGER.debug("Running %s", command)
     program, *program_args = command
     proc = await asyncio.create_subprocess_exec(

wyoming_satellite/vad.py

@@ -1,6 +1,10 @@
 """Voice activity detection."""
+import logging
+from dataclasses import dataclass
 from typing import Optional
 
+_LOGGER = logging.getLogger(__name__)
+
 
 class SileroVad:
     """Voice activity detection with silero VAD."""
@@ -31,3 +35,95 @@ class SileroVad:
             self._activation = max(0, self._activation - 1)
 
         return False
+
+
+@dataclass
+class VoiceCommandSegmenter:
+    """Segments an audio stream into voice commands."""
+
+    speech_seconds: float = 0.3
+    """Seconds of speech before voice command has started."""
+
+    silence_seconds: float = 1.0
+    """Seconds of silence after voice command has ended."""
+
+    timeout_seconds: float = 15.0
+    """Maximum number of seconds before stopping with timeout=True."""
+
+    reset_seconds: float = 1.0
+    """Seconds before reset start/stop time counters."""
+
+    in_command: bool = False
+    """True if inside voice command."""
+
+    _speech_seconds_left: float = 0.0
+    """Seconds left before considering voice command as started."""
+
+    _silence_seconds_left: float = 0.0
+    """Seconds left before considering voice command as stopped."""
+
+    _timeout_seconds_left: float = 0.0
+    """Seconds left before considering voice command timed out."""
+
+    _reset_seconds_left: float = 0.0
+    """Seconds left before resetting start/stop time counters."""
+
+    def __post_init__(self) -> None:
+        """Reset after initialization."""
+        self.reset()
+
+    def reset(self) -> None:
+        """Reset all counters and state."""
+        self._speech_seconds_left = self.speech_seconds
+        self._silence_seconds_left = self.silence_seconds
+        self._timeout_seconds_left = self.timeout_seconds
+        self._reset_seconds_left = self.reset_seconds
+        self.in_command = False
+
+    def process(self, chunk_seconds: float, is_speech: bool | None) -> bool:
+        """Process samples using external VAD.
+
+        Returns False when command is done.
+        """
+        self._timeout_seconds_left -= chunk_seconds
+        if self._timeout_seconds_left <= 0:
+            _LOGGER.warning(
+                "VAD end of speech detection timed out after %s seconds",
+                self.timeout_seconds,
+            )
+            self.reset()
+            return False
+
+        if not self.in_command:
+            if is_speech:
+                self._reset_seconds_left = self.reset_seconds
+                self._speech_seconds_left -= chunk_seconds
+                if self._speech_seconds_left <= 0:
+                    # Inside voice command
+                    self.in_command = True
+                    self._silence_seconds_left = self.silence_seconds
+                    _LOGGER.debug("Voice command started")
+            else:
+                # Reset if enough silence
+                self._reset_seconds_left -= chunk_seconds
+                if self._reset_seconds_left <= 0:
+                    self._speech_seconds_left = self.speech_seconds
+                    self._reset_seconds_left = self.reset_seconds
+        elif not is_speech:
+            # Silence in command
+            self._reset_seconds_left = self.reset_seconds
+            self._silence_seconds_left -= chunk_seconds
+            if self._silence_seconds_left <= 0:
+                # Command finished successfully
+                self.reset()
+                _LOGGER.debug("Voice command finished")
+                return False
+        else:
+            # Speech in command.
+            # Reset silence counter if enough speech.
+            self._reset_seconds_left -= chunk_seconds
+            if self._reset_seconds_left <= 0:
+                self._silence_seconds_left = self.silence_seconds
+                self._reset_seconds_left = self.reset_seconds
+
+        return True