OVOS Simple Listener

ovos-simple-listener is a lightweight alternative to ovos-dinkum-listener. It handles wake word detection, voice activity detection (VAD), and speech-to-text (STT) transcription for the Open Voice OS (OVOS) framework.

It was built to power hivemind-listener and hivemind-mic-satellite. You can also use it in place of ovos-dinkum-listener in your OVOS setup.

At around 150 lines of code, this repo is a clean reference for how to use OVOS audio plugins in your own applications.

Features

• Wake Word Detection: supports customizable wake word engines to start listening.
• Voice Activity Detection (VAD): detects silence and speech to optimize audio processing.
• Speech Recognition: uses various speech-to-text (STT) engines to transcribe audio input.
• Callback System: gives a flexible callback mechanism to handle state changes and processed audio.
• Multithreading Support: runs in a separate thread so it does not block the main application flow.

This repo is lighter than ovos-dinkum-listener, so it is also missing some features:

• Audio Transformer plugins
• Continuous Listening
• Hybrid Listening
• Recording Mode
• Sleep Mode
• Multiple WakeWords

Installation

Install ovos-simple-listener with pip:

pip install ovos-simple-listener

OVOS Usage

Run ovos_simple_listener/__main__.py in place of ovos-dinkum-listener. Plugins are selected from the default OVOS config at ~/.config/mycroft/mycroft.conf.

Library Usage

Initialize ovos-simple-listener with the components you want (microphone, STT, VAD, and wake word), as shown below.

from ovos_simple_listener import SimpleListener
from ovos_plugin_manager.microphone import OVOSMicrophoneFactory
from ovos_plugin_manager.stt import OVOSSTTFactory
from ovos_plugin_manager.vad import OVOSVADFactory
from ovos_plugin_manager.wakewords import OVOSWakeWordFactory

listener = SimpleListener(
    mic=OVOSMicrophoneFactory.create(),
    vad=OVOSVADFactory.create(),
    wakeword=OVOSWakeWordFactory.create_hotword("hey_mycroft"),
    stt=OVOSSTTFactory.create()
)

listener.run()

Callbacks

You can add your own callbacks by extending the ListenerCallbacks class. Use it to handle events such as starting a command, ending listening, processing audio, errors, and recognizing text.

from ovos_simple_listener import ListenerCallbacks

class MyCallbacks(ListenerCallbacks):
    @classmethod
    def listen_callback(cls):
        # Handle when the listener starts processing a command
        pass

    @classmethod
    def end_listen_callback(cls):
        # Handle when the listener stops processing a command
        pass

    @classmethod
    def audio_callback(cls, audio):
        # Handle processed audio data
        pass

    @classmethod
    def error_callback(cls, audio):
        # Handle STT errors
        pass

    @classmethod
    def text_callback(cls, utterance, lang):
        # Handle recognized text
        pass

Related Projects

• ovos-dinkum-listener: the full-featured production listener this repo is a lightweight alternative to.
• ovos-plugin-manager: supplies the microphone, STT, VAD, and wake word factories this repo uses.
• ovos-core: receives the recognized utterances this repo sends over the messagebus.
• hivemind-listener: a primary use case for this listener.
• hivemind-mic-satellite: a satellite device that uses this listener.

See docs/index.md for a full overview, and docs/architecture.md for the pipeline architecture in detail.

Contributing

Contributions are welcome. Open an issue or submit a pull request for any improvements or bug fixes.

Acknowledgements

• Open Voice OS for the framework and plugins.

License

This project is licensed under the terms in LICENSE.