HiveMind Voice Satellite

Mycroft Voice Satellite, connect to Mycroft HiveMind

Setup
Usage
Configuration

Install

$ pip install HiveMind-voice-sat

Usage

If host is not provided auto discovery will be used

$ HiveMind-voice-sat --help

usage: HiveMind-voice-sat [-h] [--access_key ACCESS_KEY] [--crypto_key CRYPTO_KEY] [--name NAME] [--host HOST] [--port PORT]

optional arguments:
  -h, --help            show this help message and exit
  --access_key ACCESS_KEY
                        access key
  --crypto_key CRYPTO_KEY
                        payload encryption key
  --name NAME           human readable device name
  --host HOST           HiveMind host
  --port PORT           HiveMind port number

Default values are

--access_key - "RESISTENCEisFUTILE"
--crypto_key - "resistanceISfutile"
--name - "JarbasVoiceTerminal"
--port" - 5678

Configuration

You can set the configuration at

~/.cache/json_database/HivemindVoiceSatellite.json

Otherwise default configuration will be used, check bellow for defaults

configure speech to text

{
    "lang": "en-us",
    "stt": {
        "module": "google"
    }
}

configure text to speech

{
    "lang": "en-us",
    "tts": {
        "module": "responsive_voice"
    }
}

Configure hotwords

add any number of hot words to config

hot word can be any engine (snowboy/pocketsphinx/precise)
hot word can trigger listening or not
hot word can play a sound or not
hot word can be treated as full utterance or not

{
    "hotwords": {
        "hey mycroft": {
            "module": "pocketsphinx",
            "phonemes": "HH EY . M AY K R AO F T",
            "threshold": 1e-90,
            "lang": "en-us",
            "sound": "snd/start_listening.wav",
            "listen": true
        },
        "thank you": {
            "module": "pocketsphinx",
            "phonemes": "TH AE NG K . Y UW .",
            "threshold": 0.1,
            "listen": false,
            "utterance": "thank you",
            "active": false,
            "sound": "",
            "lang": "en-us"
        },
        "wake up": {
            "module": "pocketsphinx",
            "phonemes": "W EY K . AH P",
            "threshold": 1e-20,
            "lang": "en-us"
        }
    }
}

configure listener

{
    "data_dir": "~/jarbasHiveMind/recordings",
    "listener": {
        "sample_rate": 16000,
        "channels": 1,
        "record_wake_words": false,
        "record_utterances": false,
        "phoneme_duration": 120,
        "multiplier": 1.0,
        "energy_ratio": 1.5,
        "stand_up_word": "wake up",
        "signal_folder": "/tmp/hivemind/ipc",
        "listen_sound": "snd/start_listening.wav",
        "error_sound": "snd/listening_error.mp3"
    }
}

data_dir is where recordings are saved,

{data_dir}/utterances
{data_dir}/hotwords

you can optionally set device_index

{
    "listener": {
        "device_index": 0
    }
}

or device_name, which is a name or regex pattern

{
    "listener": {
        "device_name": "respeaker"
    }
}

To trigger listening without a wakeword you can use an external signal

 touch {ipc_path}/signal/startListening

ipc_path can be set in config, default is /tmp/hivemind/ipc

Configure audio playback

you shouldn"t need to change this, a common change is replacing aplay with paplay when using pulseaudio, or use sox for everything

{
    "playback": {
        "play_wav_cmd": "aplay %1",
        "play_mp3_cmd": "mpg123 %1",
        "play_ogg_cmd": "ogg123 -q %1",
        "play_fallback_cmd": "play %1"
    }
}

flo-mic/HiveMind-voice-sat