azalea/VITS-fast-fine-tuning
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

upload files

Browse Source
This commit is contained in:
Plachta
2023-02-23 17:11:09 +08:00
parent 60bd93a26f
commit 4f09046710
1 changed files with 10 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
whisper_transcribe.py
+10 -11
View File
@@ -1,16 +1,14 @@
import whisper import whisper
import os import os
import torchaudio import torchaudio
import json
import text
lang2token = { lang2token = {
'zh': "[ZH]", 'zh': "[ZH]",
'ja': "[JA]", 'ja': "[JA]",
"en": "[EN]", "en": "[EN]",
} }
# pyopenjtalk warmup
# text._clean_text("これは困りますね", ["cjke_cleaners2"])
def transcribe_one(audio_path): def transcribe_one(audio_path):
# load audio and pad/trim it to fit 30 seconds # load audio and pad/trim it to fit 30 seconds
@@ -63,22 +61,23 @@ if __name__ == "__main__":
speaker_annos.append(save_path + "|" + str(speaker2id[speaker]) + "|" + text) speaker_annos.append(save_path + "|" + str(speaker2id[speaker]) + "|" + text)
except: except:
continue continue
import text
# clean annotation # clean annotation
cleaned_speaker_annos = []
for i, line in enumerate(speaker_annos): for i, line in enumerate(speaker_annos):
path, sid, txt = line.split("|") path, sid, txt = line.split("|")
if len(txt) > 100:
continue
cleaned_text = text._clean_text(txt, ["cjke_cleaners2"]) cleaned_text = text._clean_text(txt, ["cjke_cleaners2"])
cleaned_text += "\n" if not cleaned_text.endswith("\n") else "" cleaned_text += "\n" if not cleaned_text.endswith("\n") else ""
speaker_annos[i] = path + "|" + sid + "|" + cleaned_text cleaned_speaker_annos.append(path + "|" + sid + "|" + cleaned_text)
# write into annotation
with open("custom_character_anno.txt", 'w', encoding='utf-8') as f: with open("custom_character_anno.txt", 'w', encoding='utf-8') as f:
for line in speaker_annos: for line in cleaned_speaker_annos:
f.write(line) f.write(line)
import json
# generate new config # generate new config
with open("./configs/finetune_speaker.json", 'r', encoding='utf-8') as f: with open("./configs/finetune_speaker.json", 'r', encoding='utf-8') as f:
hps = json.load(f) hps = json.load(f)
# modify n_speakers # modify n_speakers
hps['data']["n_speakers"] = 1000 + len(speaker2id) hps['data']["n_speakers"] = 1000 + len(speaker2id)
# add speaker names # add speaker names
@@ -87,4 +86,4 @@ if __name__ == "__main__":
# save modified config # save modified config
with open("./configs/modified_finetune_speaker.json", 'w', encoding='utf-8') as f: with open("./configs/modified_finetune_speaker.json", 'w', encoding='utf-8') as f:
json.dump(hps, f, indent=2) json.dump(hps, f, indent=2)
print("finished") print("finished")