LS Maps voice guidance was generated with Piper text-to-speech (https://github.com/rhasspy/piper)
using the "en_US libritts high" voice model (speaker 552), trained on the LibriTTS dataset
(http://www.openslr.org/60/), licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/).
The phrases were written for LS Lifestyle; trimmed, levelled and encoded (mono 22 kHz Ogg Vorbis) with ffmpeg.
