ringr 0.1.4

Sound event detection system

ringr

Sound event detection system based on the open-source cross-platform PortAudio API.

It supports multiple notification backends, and it has been designed to run on low-specs and inexpensive hardware like the lowest range of raspberry products (RPi 1, RPi Zero, RPi Zero 2, etc).

Use cases

Some interesting use cases where to use ringr to automate your smart home and help disabled people:

• Intercom and doorbell detection
• Detection of the end-of-program audible warning of some appliances such as washing machines, clothes dryers or dishwashers
• Baby crying detection
• Dog bark detection

Design considerations and constraints

This has born as a personal project for a very concrete use case that can be extrapolated to multiple other contexts and needs.

Some of my personal requirements and constraints include:

• It must run on an old Raspberry Pi 1 that I have stored in a drawer and unused for some many years to give it a new life.
• It must run with very cheap and low quality microphones.
• It must be able to detect the intercom and doorbell notification sounds without false positives warnings.
• It must notify states to my personal Home Assistant installation and provide support to add more notification backends in the future.

More information about the concept and design of ringr in this post on my personal blog.

Installation

On all systems, install ringr by using pip:

pip install ringr

Systemd

Copy the file ringr.service available in this repository to /etc/systemd/system/ringr.service

Edit the service and modify the application path, the user and the group if needed.

Reload the systemd daemon to load the new service by executing systemctl daemon-reload.

Start the service with systemctl start ringr.service and enable it if you want to run it automatically at startup: systemctl enable ringr.service

Docker

Alternatively to a native installation you can use Docker.

A Dockerfile is provided to build a Docker image for ringr.

docker build -t ringr .

No volumes are needed as ringr is a stateless service.

Remember you need to add the host sound device to your container with --device

docker run -d \
  --name=ringr \
  --device=/dev/snd:/dev/snd
  -e TZ=Europe/Madrid \
  -e PUID=1000 \
  -e GUID=1000 \
  -e RINGR_DETECTOR_DEVICE='1' \
  -e RINGR_DETECTOR_THRESHOLD='65' \
  -e RINGR_DETECTOR_PEAK_DURATION='0.8' \
  -e RINGR_DETECTOR_FREQUENCY='1000' \
  -e RINGR_DETECTOR_ACCEPTANCE_RATIO='90' \
  -e RINGR_DETECTOR_GAIN='200' \
  -e RINGR_DETECTOR_LATENCY='0.1' \
  -e RINGR_DETECTOR_COOLDOWN='15' \
  -e RINGR_NOTIFIER_TYPE='ha' \
  -e RINGR_NOTIFIER_MQTT_HOST='10.10.0.50' \
  -e RINGR_NOTIFIER_USER='ringr' \
  -e RINGR_NOTIFIER_PASS='secret' \
  --restart unless-stopped \
  ringr:latest

docker-compose

Example of a docker-compose file:

version: '3'
services:
  ringr:
    container_name: ringr
    image: ringr:latest
    devices:
      - '/dev/snd:/dev/snd'
    environment:
      TZ: 'Europe/Madrid'
      PUID: '1000'
      GUID: '1000'
      RINGR_DETECTOR_DEVICE: '1'
      RINGR_DETECTOR_THRESHOLD: '65'
      RINGR_DETECTOR_PEAK_DURATION: '0.8'
      RINGR_DETECTOR_FREQUENCY: '1000'
      RINGR_DETECTOR_ACCEPTANCE_RATIO: '90'
      RINGR_DETECTOR_GAIN: '200'
      RINGR_DETECTOR_LATENCY: '0.1'
      RINGR_DETECTOR_COOLDOWN: '15'
      RINGR_NOTIFIER_TYPE: 'ha'
      RINGR_NOTIFIER_MQTT_HOST: '10.10.0.50'
      RINGR_NOTIFIER_USER: 'ringr'
      RINGR_NOTIFIER_PASS: 'secret'
    restart: unless-stopped

Usage

Use the ringr command to launch the application.

It supports the following optional parameters:

• -c, --conf: configuration file. By default it uses /etc/ringr/ringr.conf
• -v, --verbose: configure the log level of the logger in debug level

Configuration

ringr can be configured through a configuration file or with environment variables, useful if you run it within a docker container.

The expected path to the configuration file is /etc/ringr/ringr.conf.

Detector

The configuration of the detector is defined inside the [detector] section of the configuration files.

It supports the following options:

device

Option	Environment variable	Data type	Unit	Default
device	RINGR_DETECTOR_DEVICE	integer

Index of the input sound device.

You can get the list of PortAudio sound devices and their index with the following command:

$ python -m sounddevice
  0 sof-hda-dsp: - (hw:1,0), ALSA (2 in, 2 out)
  1 sof-hda-dsp: - (hw:1,4), ALSA (0 in, 2 out)
  2 sof-hda-dsp: - (hw:1,5), ALSA (0 in, 2 out)
  3 sof-hda-dsp: - (hw:1,6), ALSA (2 in, 0 out)
  4 sof-hda-dsp: - (hw:1,7), ALSA (2 in, 0 out)
  5 pulse, ALSA (32 in, 32 out)
* 6 default, ALSA (32 in, 32 out)

threshold

Option	Environment variable	Data type	Unit	Default
threshold	RINGR_DETECTOR_THRESHOLD	float	%

Relative amplitude threshold for the event detection. It must be a value in the range [0,100].

peak_duration

Option	Environment variable	Data type	Unit	Default
peak_duration	RINGR_DETECTOR_PEAK_DURATION	float	secs.

Duration of the signal over the threshold amplitude before to be considered an event.

frequency

Option	Environment variable	Data type	Unit	Default
frequency	RINGR_DETECTOR_FREQUENCY	int	Hz

Frequency where to analyze the event.

The event will be analyzed inside a frequency range where the highest frequency will be the closest possible frequency to the desired frequency based on the number of frequency bins used.

frequency_bins

Option	Environment variable	Data type	Unit	Default
frequency_bins	RINGR_DETECTOR_FREQUENCY_BINS	int		256

Number of frequency bins in which divide the range of audible frequencies.

The detector will analyze the frequency bin whose highest frequency is closest to the desired frequency.

For example, with frequency = 1000 and frequency_bins = 256, the detector will analyze the range between 951 Hz and 1037 Hz.

acceptance_ratio

Option	Environment variable	Data type	Unit	Default
acceptance_ratio	RINGR_DETECTOR_ACCEPTANCE_RATIO	float	%	100

Number of samples that must be above the amplitude threshold in the desired frequency range during the peak_duration time.

gain

Option	Environment variable	Data type	Unit	Default
gain	RINGR_DETECTOR_GAIN	int		0

Some input devices may capture very low signals. Use this parameter to boost them to have more control on its amplitude.

latency

Option	Environment variable	Data type	Unit	Default
latency	RINGR_DETECTOR_LATENCY	float	secs.	high

Input latency of capture.

Higher values will prevent input overflow errors that will discard samples and produce more robust and predictable results.

By default, it will use the predefined high input latency of your input sound device.

You can query that value using the following command replacing the argument in the query_devices method by the device index of your device:

$ python -c 'import sounddevice as sd; print(sd.query_devices(1))'
{'name': 'default', 'index': 11, 'hostapi': 0, 'max_input_channels': 32, 'max_output_channels': 32, 'default_low_input_latency': 0.008684807256235827, 'default_low_output_latency': 0.008684807256235827, 'default_high_input_latency': 0.034807256235827665, 'default_high_output_latency': 0.034807256235827665, 'default_samplerate': 44100.0}

cooldown

Option	Environment variable	Data type	Unit	Default
cooldown	RINGR_DETECTOR_COOLDOWN	float	secs.	10

Cooldown time in seconds after a detection. Any sample during this period will be discarded and will not be analyzed.

block_duration

Option	Environment variable	Data type	Unit	Default
block_duration	RINGR_DETECTOR_BLOCK_DURATION	float	millisecs.	50

Duration of the sample to analyze.

Higher values may provoke input overflow errors.

log_analysis

Option	Environment variable	Data type	Unit	Default
log_analysis	RINGR_DETECTOR_LOG_ANALYSIS	bool		False

Verbose output of analysis. Use with log level = debug

Notification backends

The configuration of the chosen notification backend is defined inside the [notifier] section of the configuration file.

All notification backends share one option: type which specifies the notification backend to use. The corresponding environment variable for this option is RINGR_NOTIFIER_TYPE

Available notification backends:

Name	Type	Description
Home Assistant	ha	Auto-discoverable MQTT device for Home Assistant
Telegram	telegram	Telegram Bot

Home Assistant

Use type: ha

Option	Environment variable	Data type	Default	Description
device_id	RINGR_NOTIFIER_DEVICE_ID	str	ringr_01	Unique device id. Override this value if you run multiple instances of ringr
device_name	RINGR_NOTIFIER_DEVICE_NAME	str	ringr 01	Device name. Override this value if you run multiple instances of ringr or to use a custom name within Home Assistant
mqtt_host	RINGR_NOTIFIER_MQTT_HOST	str		Host of the MQTT broker
mqtt_port	RINGR_NOTIFIER_MQTT_PORT	int	1883	Port of the MQTT broker
mqtt_user	RINGR_NOTIFIER_MQTT_USER	str		Optional username to authenticate with the MQTT broker. Use it together with mqtt_pass
mqtt_pass	RINGR_NOTIFIER_MQTT_PASS	str		Optional password to authenticate with the MQTT broker. Use it together with mqtt_user
mqtt_client_id	RINGR_NOTIFIER_MQTT_CLIENT_ID	str	ringr_01	Client id of the MQTT client. Override this value if you run multiple instances of ringr
mqtt_qos	RINGR_NOTIFIER_MQTT_QOS	int	1	MQTT QoS of published messages

Telegram

Use type: telegram

Option	Environment variable	Data type	Default	Description
api_token	RINGR_NOTIFIER_API_TOKEN	str		Telegram Bot API Token
chat_id	RINGR_NOTIFIER_CHAT_ID	str		Telegram Chat ID
message	RINGR_NOTIFIER_MESSAGE	str	Event detected	Message to send where an event is detected

Full example

Configuration file:

[detector]
device: 1
threshold: 65
peak_duration: 0.8
frequency: 1000
acceptance_ratio: 90
gain: 200
latency: 0.1
cooldown: 15

[notifier]
type: ha
mqtt_host: 10.10.0.50
mqtt_user: ringr
mqtt_pass: secret

Or, by using environment variables:

RINGR_DETECTOR_DEVICE='1'
RINGR_DETECTOR_THRESHOLD='65'
RINGR_DETECTOR_PEAK_DURATION='0.8'
RINGR_DETECTOR_FREQUENCY='1000'
RINGR_DETECTOR_ACCEPTANCE_RATIO='90'
RINGR_DETECTOR_GAIN='200'
RINGR_DETECTOR_LATENCY='0.1'
RINGR_DETECTOR_COOLDOWN='15'

RINGR_NOTIFIER_TYPE='ha'
RINGR_NOTIFIER_MQTT_HOST='10.10.0.50'
RINGR_NOTIFIER_USER='ringr'
RINGR_NOTIFIER_PASS='secret'