ecker/tortoise-tts
4
6
Fork 15
Code Issues 9 Pull Requests 1 Packages Projects Releases Wiki Activity

Upgrade of text splitting

Browse Source 
Upgrade of text splitting in read.py. Now text splits across '!' '?' ';', and double quotations ("") keeps together. 

Example:

Original:
They're often cooked in an unusual way, too. What they are doing? Take carpa regina in porchetta, one of the lake's signature dishes. "Carp baked like porchetta!" (herb-roasted pork) takes one of the lake's biggest fishes, slathers it in strong herbs, and roasts it - just as is done with Italy's classic meat, porchetta.

Splitted:
"I come to Trasimeno for the food - because it's like nowhere else in Italy," says Veronica Grechi, a B&B owner from Florence, and regular visitor to the lake.
["They're often cooked in an unusual way, too.", 'What they are doing?', "Take carpa regina in porchetta, one of the lake's signature dishes.", '"Carp baked like porchetta!" (herb-roasted pork) takes one of the lake\'s biggest fishes, slathers it in strong herbs, and roasts it - just as is done with Italy\'s classic meat, porchetta.', '"I come to Trasimeno for the food - because it\'s like nowhere else in Italy," says Veronica Grechi, a B&B owner from Florence, and regular visitor to the lake.']
This commit is contained in:
faad3 2022-05-13 12:19:32 +03:00 committed by GitHub
parent 945bd88f21
commit 90916da81a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
tortoise/read.py
View File

@ -1,5 +1,6 @@
import argparse
import os
import re
import torch
import torchaudio
@ -9,8 +10,8 @@ from tortoise.utils.audio import load_audio, get_voices, load_voices
def split_and_recombine_text(texts, desired_length=200, max_len=300):
# TODO: also split across '!' and '?'. Attempt to keep quotations together.
texts = [s.strip() + "." for s in texts.split('.')]
split = re.split(r'([\!\?\;]|\.{1,3})+(?=(?:(?:[^\"\n]*\"){2})*[^\"\n]*$)',texts, flags=re.MULTILINE)
texts = [split[i-1].strip()+split[i].strip() for i in range(1,len(split),2)]
i = 0
while i < len(texts):