A Simple Speechhandler Plugin

In this section we are going to work through the case of developing a simple speechhandler plugin to give Naomi the skill to implement a one to five second timer. In operation it will allow the user to verbally specify the number of seconds to use and then Naomi will announce the timer start, then after the elapse time it will announce the timer finish.

Note that the code developed below has actually two functions that will be used by the TTI and Speechhandler modules to "decode" input and implement the output actions.

Structure of the Plugin

At the top level a speechhandler plugin is a directory containing at minimum of three files. For user plugins this directer would be placed under the "hidden" directory ~/.config/naomi/plugin/speechhandler. For the purposes of this tutorial name the directory with your files "simple_timer".

__init.py__ File

This file contains the import instruction for loading the plugin. In this case it is simply two lines. The value "s_timer" is the name of the Python file containing the code and "SimpleTimerPlugin" is the name of the module implementing the timer.

## -*- coding: utf-8 -*-
from .s_timer import SimpleTimerPlugin

plugin.info File

Contains metadata about the plugin. In our case the contents should be fairly obvious. The "name", "Version" (version scheme is arbitrary), and "Description" are what appear when you run Naomi with the --list-active-plugins. Other values are simply informational and not operationally effecting.

Name = Simple Timer
Version = 1.0.0
License = MIT
URL = https://github.com/mcoyne1948/simple_timer
Description = Very basic elapse timer

Name = Malcolm Coyne
URL = https://malcolmc.ca/

s_timer.py File

Is the actual Python code implementing the plugin and has basically three main divisions. The first is initialization statements. The second is a Naomi readable description to assist it in recognizing the intent of this plugin in the incoming stream of spoken words. Finally, the last division implements the handler for when this intent is triggered.


In the initialization we need to import the Python time module that provides the sleep method used to measure our timer elapse times. We also need the plugin module from Naomi and finally we need to define the class for our module.

import time
from naomi import plugin
class SimpleTimerPlugin(plugin.SpeechhandlerPlugin):

Intent Function

This function provides the framework for interpreting the input word stream and identifying the timer intent is to be activated. The code can be seen below.

def intents(self):
        return {
            'TimerIntent': {
                'locale': {
                    'en-US': {
                        'keywords': {
                            'NumberKeyword': [
                        'templates': [
                            "SET TIMER {NumberKeyword} SECONDS",
                            "SET {NumberKeyword} SECOND TIMER"
                'action': self.handle

The en-US indicates the start of a language specific section and there can be multiple sections of this type in a single file for different languages. Within the language section there is an optional keywords subsection which define the options to be considered for one of the input ranges being evaluated. The second subsection, templates, define the lexical structures that the input stream is going to be tested against to determine if it specifies the timer intent.

Handler Function

The handler defines the action(s) to be taken once the timer intent has been identified. It has four major section; the function specifier, dictionary initialize, inputs test and code to perform the timer function. The first line shown below simply defines the function name and the inputs which in this case are the intent information and the mic voice I/O functions.

def handle(self, intent, mic):

The next section initialize some function variables. In this case a NUMBERS dictionary for decoding numeric values from textual numbers is define, and default for the ERROR and NUMBER variable are set.

        NUMBERS = {
            'ONE' :    1,
            'TWO' :    2,
            'THREE' :  3,
            'FOUR' :   4,
            'FIVE' :   5
        ERROR = ''
        NUMBER ='ONE'

The next section first extracts the number identified by the NumberKeyword in the intent stream. If this is missing or can't be resolve it sets an appropriate error. Next it tries to lookup the NUMBER in the NUMBERS dictionary. If this should fail then another error is set.

            NUMBER = intent['matches']['NumberKeyword'][0]
            # print (" Number is: " + NUMBER)      # For debug
        except KeyError:
            ERROR = "Could not understand input. Please try again!"
            ETimer = NUMBERS[NUMBER]
        except KeyError:  
            ERROR = "Timer input out of range!"

The final sections takes the appropriate action for the plugin (skill). If an error has been detect Naomi announces it otherwise it proceed with the timer function. The timer first announces the start, then uses the sleep function to time the elapse implementing a delay and finishes by announcing the end.

        if ERROR != '':
            mic.say(self.gettext("ERROR: %s!" % ERROR))
        # Run elapse timer   
            mic.say(self.gettext("Start " + NUMBER  + " second timer!"))
            mic.say(self.gettext(NUMBER + " second timer finished!")

Test and Debug

Once the plugin directory with its three files have been created it needs to be place in the ~/.config/naomi/plugin/speechhandler directory. This is all that is required to make your plugin available to be loaded into Naomi.

To check if the plugin will load into Naomi enter the following: (The --debug piece is optional but will be useful for debugging if the plugin does not load.)

naomi --list-active-plugins --debug |& tee ~/naomi.log

This gives an alphabetic listing of all installed plugins so you can check if ours is in the list. If not in the list you can open the naomi.log file (it should be in your home directory) and check for errors in our plugin code. Search for "simple_timer" which should bring up a WARNING that your plugin has been "skipped". Following this line should be a traceback that can help with identifying the error in the plugin code.

Fix your code and rerun the check process. Once the plugin appears in the list you can test it by running naomi without options and when prompted speak "Set five second timer!"

If after reasonable effort you are unable to resolve your issue help may be available from the sources in the following section.

Getting Help With Plugin Development

A good method to start figuring out how to create a particular plugin is to see if in the list of installed plugins in the ~/Naomi/plugins/speechhandler directory or the Naomi Plugin Exchange (NPE) if there is one along the lines you are thinking and examine its code to see if that helps. However, if after taking a crack at it should you run into problems you have been unable to work through you can also consult the Naomi community forum pages or use the Discord community.