Model
Not loadedThe 35 MB model is downloaded once and then kept on this device. After that it works offline.
0.60×1.60×
Examples
Generated audio appears here.
What the model actually saw
Your text is turned into phonemes by eSpeak NG, then rewritten into the spelling this model was trained on. Feeding it raw eSpeak phonemes roughly triples the word error rate, so this step is not optional.