Announcements for the Lyrion Media Server

LMS Announce is a plugin for playing announcements / notifications (audio / speech) using LMS.

Features

  • Web API (using HTTP GET, i.e. calling an URL)
  • Can play an audio file / URL, a playlist or generate speech (using a Text to Speech (TTS) engine installed on the LMS server)
  • Pauses current playlist, plays announcement, resumes playlist

Scenarios

  • Home automation
  • Phone (e.g. using IFTTT or Automate)
  • Anything that can call an URL with parameters (wget from a shell script etc.)

Status

  • Beta quality
    • untested with any sort of streaming
    • little error handling
  • Tested on Linux
  • May work on MacOS (provided the TTS profiles are adapted, see below)
  • Most likely won't fully work on Windows, except maybe the playlist and/or file mode

Installation

In the LMS standard web interface in Settings/Plugins, add https://www.nexus0.net/pub/sw/lmsannounce/repo.xml to the list of Additional Repositories and activate the plugin.

Usage Instructions

The plugin provides it's API at http://lmsserver:9000/plugins/LMSannounce/js.html. All calls require the cmd and playerid parameter, plus, depending on the command, other parameters as well.

Note: As always, parameter values need to be URL encoded.
curl can do this automatically:

curl -G  'http://lmsserver:9000/plugins/LMSannounce/js.html' --data-urlencode "cmd=announceTTS" --data-urlencode "ttsprofile=mimic-ap" --data-urlencode "playerid=00:00:00:00:00:00" --data-urlencode "text=new message"
Announce (playlist)
Parameter Value
cmd announcePlaylist
playerid ID of player (MAC address)
playlist Name of playlist
forcewake true / false (force player to wake up (default: true))
volume 0-100

Example: http://lmsserver:9000/plugins/LMSannounce/js.html?cmd=announcePlaylist&playlist=somepls&playerid=00%3A00%3A00%3A00%3A00%3A00

Announce (file or URL)
Parameter Value
cmd announceFile
playerid ID of player (MAC address)
file Full (absolute) path of an audio file or an URL
forcewake true / false (force player to wake up (default: true))
volume 0-100

Example: http://lmsserver:9000/plugins/LMSannounce/js.html?cmd=announceFile&file=%2Fdata%2Fmusic%2Fsoundeffects%2Fring.mp3&playerid=00%3A00%3A00%3A00%3A00%3A00

Announce (TTS)
Parameter Value
cmd announceTTS
playerid ID of player (MAC address)
ttsprofile select TTS profile (see below)
text Text to speak
forcewake true / false (force player to wake up (default: true))
volume 0-100

Example: http://lmsserver:9000/plugins/LMSannounce/js.html?cmd=announceTTS&ttsprofile=espeak-en-m&text=new%20message&playerid=00%3A00%3A00%3A00%3A00%3A00

Text to Speech

The plugin needs a TTS engine installed on the server on which LMS is running (or you could use a cloud service, e.g. AWS Polly or Google TTS). This engine must be able to convert text to an audio file using a CLI command.

Some options are:

  • espeak (supported, small and fast, robotic voice)
  • mimic (supported, good quality and fairly natural voices)
  • SVOX pico
  • festival, flite, ... (not tested, would most likely work)
  • Larynx (untested, but very interesting. Reported as working by 3rd party.)
  • coqui-ai/TTS (untested, but very interesting)
  • gTTS (Google TTS CLI tool. Reported as working by 3rd party.)

Some of them should be available as a package in most Linux distributions.

TTS profiles

A profile defines which TTS engine to use, and it's parameters. There are a number of defaults, and it's possible to define custom profiles as well.

Profile Description
espeak-en-m espeak TTS, english, male
espeak-en-f espeak TTS, english, female
espeak-de-m espeak TTS, german, male
espeak-de-f espeak TTS, german, female
mimic-slt mimic TTS, SLT voice
mimic-ap mimic TTS, AP voice
Custom TTS profiles

To support other TTS engines / voices, create a file named la-ttsprofiles.json in /etc or /usr/local/etc containing the definitions as a JSON object.

Example (download): 

    {
        "espeak-fr-f": {
            "cmd": "/usr/bin/espeak",
            "output": "-w ",
            "options": "-vfr+f3",
            "text": ""
        },
        "mimic-rms": {
            "cmd": "/usr/bin/mimic",
            "output": "-o ",
            "options": "-voice rms",
            "text": "-t"
        }
    }
  • cmd: the command to execute
  • output: parameter to specify the output file (wav format)
  • options: parameter to specify any other options to pass to the command
  • text: parameter to specify the text to convert

The profiles can then be used as usual: http://lmsserver:9000/plugins/LMSannounce/js.html?cmd=announceTTS&ttsprofile=espeak-fr-f&text=ordinateur%20deux%20trois&playerid=00%3A00%3A00%3A00%3A00%3A00

Tips & Tricks

Pre-roll / post-roll sound
  • for static announcements, use a playlist
  • for speech, install sox and create a shell script:
#!/bin/bash

SOUND="/path/to/beep.flac" 
while getopts "o:t:" opt; do
        case $opt in
                o)
                DESTFILE="$OPTARG"
                ;;
                t)
                TEXT="$OPTARG"
                ;;
        esac
done
mimic -o "${DESTFILE}.tmp" -voice slt -t "$TEXT"
sox "${DESTFILE}.tmp" -t sox - channels 2 rate 44100 | sox "$SOUND" -t sox - "$DESTFILE" 
rm  "${DESTFILE}.tmp"

and use it as a TTS engine:

    "mimic-beep": {
        "cmd": "/path/to/script.sh",
        "output": "-o ",
        "options": "",
        "text": "-t"
    }
Using a remote TTS

If you cannot install a TTS on the LMS server, you can run a non-local one using ssh:

#!/bin/bash

while getopts "o:t:" opt; do
        case $opt in
                o)
                DESTFILE="$OPTARG"
                ;;
                t)
                TEXT="$OPTARG"
                ;;
        esac
done
ssh user@otherserver "/usr/bin/mimic -o /dev/stdout -voice slt -t \"$TEXT\"" > "${DESTFILE}"
Sending text interactively
$  function LMS_announce { curl -G -s -o /dev/null  'http://lmsserver:9000/plugins/LMSannounce/js.html' --data-urlencode "cmd=announceTTS" --data-urlencode "ttsprofile=mimic-slt" --data-urlencode "playerid=00:00:00:00:00:00" --data-urlencode "text=$1"; }
$  LMS_announce "hello world"

FAQ

Adding the plugin repository to LMS results in some errors / doesn't work. Why?
Make sure your LMS installation is SSL-enabled (e.g. on *nix, the perl modules Net::SSLeay and IO:Socket:SSL need to be installed)
What does 'Error Plugins::LMSannounce::Plugin::ANON (306) playlist creation error' mean?
Check if playlists can actually be created on your LMS installation. Check if a playlist called announceTTS exists and delete it.
When using TTS, there is only silence / the call returns "error n while executing TTS". What now?
First, check from the shell if your TTS engine actually works (using the account LMS runs as, e.g. squeezeboxserver). After that, switch logging to debug for the plugin (LMS Web interface/Settings/Advanced/Logging/plugin.lmsannounce). Search the server log for lines like announceTTS: command: /usr/bin/mimic -o /tmp/annTTS.wav -voice awb -t "message" and check if the command is correct and works from a regular shell (delete /tmp/annTTS.wav when done).

License

GNU GPLv3

Reporting Issues

Use the discussion thread in the LMS forum.