Components for Home Assistant

macOS Text-to-Speech and Media Player

This component uses the locally installed, high quality text-to-speech (TTS) engine in macOS and the locally installed afplay macOS utility to play media files on the selected sound output source.

Prerequisites

As of Mac OS X 10.6, macOS requires that the speech synthesizer is invoked inside a user name space - otherwise, the say process will just hang forever because the system cannot create a speechsynthesizerd process.

Since Home Assistant itself will run as a daemon (detached from the user), the macOS TTS component requires the installation of a helper program. The easiest way to do this is to install using HomeBrew (https://brew.sh):

brew install reattach-to-user-namespace

This installs a small executable into /usr/local/bin, aptly named reattach-to-user-namespace. See https://github.com/ChrisJohnsen/tmux-MacOSX-pasteboard for technical details.

Component Installation

Copy custom_components/macos_tts to custom_components/macos_tts inside your Home Assistant configuration folder.

Enable and configure in configuration.yaml:

tts:
  - platform: macos_tts
    voice: 'Ava'

media_player:
  - platform: macos_tts

Available voices are listed in System Preferences (Accessibility > Speech as of High Sierra), or you can use the say command:

$ say -v'?'
Amelie              fr_CA    # Bonjour, je m’appelle Amelie. Je suis une voix canadienne.
Ava                 en_US    # Hello, my name is Ava. I am an American-English voice.
Daniel              en_GB    # Hello, my name is Daniel. I am a British-English voice.
Diego               es_AR    # Hola, me llamo Diego y soy una voz española.
...

Make sure you download at least one high-quality voice using System Preferences (possibly listed as "enhanced" or "premium"). Alex, Ava, and Samantha are good choices.

Note: While it's possible to configure a language such as en-US, the language setting is ignored. Select a matching voice instead.

Component Test

Use the developer tools - services in the web interface to call the service tts.macos_tts_say with service data

{"message": "Hello world, can you hear me now?"}

If the Mac's default speaker is on and the media player component is working, you should hear the synthesized voice. If you have not yet configured a media player, and the log level is set to allow it, Home Assistant will add a log entry

2017-12-29 23:55:02 WARNING (MainThread) [homeassistant.core] Unable to find service ...

In any case, the synthesized speech can be found in the cache directory tts, for example:

-rw-r--r--  1 _ha _ha 44552 Dec 29 23:55 cafebabedeadbeef0badabbababafeca_en-us_-_macos_tts.m4a

The macOS TTS component saves synthesized speech in m4a (MPEG-4 Audio) format. Not only does this save disk space, but it's one of the few formats both Home Assistant's TTS and macOS understand.

You can play these files using afplay.

binary_sensor.httpserver

This component starts a local http web server and waits for connections. When a client connects with a GET request, the URL is parsed and a binary sensor is triggered.

Component Installation

Copy custom_components/httpserver to custom_components/httpserver in your Home Assistant configuration folder.

Component Configuration

In configuration.yaml, add the new httpserver binary_sensor platform. Under sensors, add one sensor for each HTTP URL path you wish to respond to. Each of these will create one binary sensor in Home Assistant.

The listen_port is an available local port.

Per-Sensor Configuration

Examples

binary_sensor:
  - platform: httpserver
    listen_port: 8130
    sensors:
      foscam_motion:
        set_path: "/motion/garage"
        friendly_name: "Garage Motion (self-resetting)"
        device_class: motion
        reset_delay: 20
      toggle_sensor:
        set_path: "/sensor/toggle"
        friendly_name: "Toggle Sensor"
        device_class: motion
        initial_state: True
      set_reset_sensor:
        set_path: "/sensor/set/1"
        reset_path: "/sensor/set/0"
        friendly_name: "Set-Reset Sensor"
        device_class: motion

Restart Home Assistant.

Component Test

Use curl to trigger the binary sensor. If the Home Assistant server IP is 10.11.12.100, then the test command for the self-resetting sensor shown in the example would be:

$ curl http://10.11.12.100:8130/motion/garage

Since Home Assistant will return a "200" (OK) HTTP status, curl will not print anything on success. curl will return an error message however in case of invalid IP:port or a non-configured path.

Example: Foscam FI8910W Motion Sensor

Foscam FI8910W can post http requests when it detects motion. Note: This feature is not available on newer cameras.

The following describes cameras following the API in IP Camera CGI V1.27.

The camera does not post another request when the sensor clears, so we need to configure a self-resetting sensor.

To get the current settings of your Foscam (in the example, the camera is at address 10.11.12.13 port 80):

$ curl "http://10.11.12.13:80/get_params.cgi?user=USER&pwd=PASSWORD"

...
var alarm_motion_armed=0;
var alarm_http=0;
var alarm_http_url=;
...

Pick a free local port number on the Home Assistant system. In this example, we're using port 8130 which is not otherwise used. HA is at IP 10.11.12.100.

Configure and Test the Foscam

Enable the motion sensor (which could also be done using the camera's web UI), and set the parameters http and http_url (these are hidden in the camera's web UI) using curl:

$ curl "http://10.11.12.13:80/set_alarm.cgi?user=USER&pwd=PASSWORD&motion_armed=1&http=1&http_url=http://10.11.12.100:8130/motion/garage"
ok.

Other options such as sensitivity and motion compensation can be set in the web UI.

You can test whether the motion alarms are working without setting up a httpserver binary_sensor. To do so, start a local http server on the PORT you specified in the camera's http_url. This is easiest using nc:

$ nc -l 8130

Then, get the camera to trigger and you should see:

GET /motion/garage HTTP/1.0
HOST: 10.11.12.100:8130
User-Agent: myclient/1.0 me@null.net