Simple TTS library for MicroPython that works offline

Tis library is a MicroPython port of concatanative diphone-based speach synthesizer. It uses lexicon database to convert words to phonemes, which was converted from CMU Pronouncing Dictionary (License). Diphones database was converted from diphone collection by Alan W Black and Kevin Lenzo (License).

1. Paste folowing code to REPL console to download library

   import mip
   mip.install("github:Voinic/microtts")

2. Copy lexicon.db and any of diphones.db and diphones_lq.db to SD-card or internal flash (if your have enough space)

Use diphones_lq.db contains recordings with IMA ADPCM compression. You can use it instead of diphones.db if your memory space is limitted. Note that sound quality will decrease.

from utts import Utterance, Synth

LEXICON_DB = "/sd/lexicon.db"

DIPHONES_DB = "/sd/diphones.db"
DB_COMPRESSED = False

CROSSFADE = 0.025

TEXT = "This is a test"

utterance = Utterance(LEXICON_DB)
synth = Synth(DIPHONES_DB, DB_COMPRESSED)

utterance.process(TEXT)
diphones = utterance.get_diphones()

synth.synthesize(diphones, CROSSFADE)
audio = synth.get_audio()

• examples/save_to_wav.py - Converts given text to speach and saves result into WAV file.
• examples/direct_playback.py - Prompts user for input text and sends output audio to I2S DAC.

/db folder contains code and input files that was used for databases creation. Run this code using micropython interpreter, not regular python (I used Unix port of micropython).