scalable multiple tts识别
plan1:
-
collect single audio files from multiple tts service
-
concat single audio file into 8 characters audio files
-
convert audio to spectrogram using sox command
-
use train.py to train the model
plan2:
-
collect single audio files from multiple tts service
-
train model for single character
-
split test audio with 8 characters into single character file based on silence
-
use classify.py to recognize single character
-
combine the results from single character