I’m wondering if anyone knows of a TTS on HF that splits the output audio into separate audio files based on one audio file for each line in a text file.
1 Like
I have seen several models that can synthesize long texts by dividing them into chunks, but if you want to divide the text at specific points, it is more reliable to preprocess the text yourself and pass it to the TTS model as a batch. In this case, you can write a loop to divide each line and pass it to the model.
Speaker (voice) consistency may be an issue, but models such as Dia may be able to guarantee this to some extent. (I have not tried it myself, though.)