[TTS]Cantonese FastSpeech2 e2e infer, test=tts (#2927)
HuangLiangJie
2 years ago
committed by
GitHub
parent
004a4d6096
commit
1af9bd47d9
@ -0,0 +1,53 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
|
||||||
|
config_path=$1
|
||||||
|
train_output_path=$2
|
||||||
|
ckpt_name=$3
|
||||||
|
|
||||||
|
stage=0
|
||||||
|
stop_stage=0
|
||||||
|
|
||||||
|
# pwgan
|
||||||
|
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
|
||||||
|
FLAGS_allocator_strategy=naive_best_fit \
|
||||||
|
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||||
|
python3 ${BIN_DIR}/../synthesize_e2e.py \
|
||||||
|
--am=fastspeech2_canton \
|
||||||
|
--am_config=${config_path} \
|
||||||
|
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||||
|
--am_stat=dump/train/speech_stats.npy \
|
||||||
|
--voc=pwgan_aishell3 \
|
||||||
|
--voc_config=pwg_aishell3_ckpt_0.5/default.yaml \
|
||||||
|
--voc_ckpt=pwg_aishell3_ckpt_0.5/snapshot_iter_1000000.pdz \
|
||||||
|
--voc_stat=pwg_aishell3_ckpt_0.5/feats_stats.npy \
|
||||||
|
--lang=canton \
|
||||||
|
--text=${BIN_DIR}/../sentences_canton.txt \
|
||||||
|
--output_dir=${train_output_path}/test_e2e \
|
||||||
|
--phones_dict=dump/phone_id_map.txt \
|
||||||
|
--speaker_dict=dump/speaker_id_map.txt \
|
||||||
|
--spk_id=0 \
|
||||||
|
--inference_dir=${train_output_path}/inference
|
||||||
|
fi
|
||||||
|
|
||||||
|
# hifigan
|
||||||
|
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
|
||||||
|
echo "in hifigan syn_e2e"
|
||||||
|
FLAGS_allocator_strategy=naive_best_fit \
|
||||||
|
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||||
|
python3 ${BIN_DIR}/../synthesize_e2e.py \
|
||||||
|
--am=fastspeech2_canton \
|
||||||
|
--am_config=${config_path} \
|
||||||
|
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||||
|
--am_stat=dump/train/speech_stats.npy \
|
||||||
|
--voc=hifigan_aishell3 \
|
||||||
|
--voc_config=hifigan_aishell3_ckpt_0.2.0/default.yaml \
|
||||||
|
--voc_ckpt=hifigan_aishell3_ckpt_0.2.0/snapshot_iter_2500000.pdz \
|
||||||
|
--voc_stat=hifigan_aishell3_ckpt_0.2.0/feats_stats.npy \
|
||||||
|
--lang=canton \
|
||||||
|
--text=${BIN_DIR}/../sentences_canton.txt \
|
||||||
|
--output_dir=${train_output_path}/test_e2e \
|
||||||
|
--phones_dict=dump/phone_id_map.txt \
|
||||||
|
--speaker_dict=dump/speaker_id_map.txt \
|
||||||
|
--spk_id=0 \
|
||||||
|
--inference_dir=${train_output_path}/inference
|
||||||
|
fi
|
||||||
@ -0,0 +1,7 @@
|
|||||||
|
001 白云山爬过一次嘅，好远啊，爬上去都成两个钟
|
||||||
|
002 睇书咯，番屋企，而家好多人好少睇书噶喎
|
||||||
|
003 因为如果唔考试嘅话，工资好低噶
|
||||||
|
004 冇固定噶，你中意休边日就边日噶
|
||||||
|
005 即系太迟嘅话咧，落班太迟嘅话就喺出边食啲咯
|
||||||
|
006 是非有公理，慎言莫冒犯别人
|
||||||
|
007 遇上冷风雨，休太认真
|
||||||
@ -0,0 +1,106 @@
|
|||||||
|
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
||||||
|
#
|
||||||
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||||
|
# you may not use this file except in compliance with the License.
|
||||||
|
# You may obtain a copy of the License at
|
||||||
|
#
|
||||||
|
# http://www.apache.org/licenses/LICENSE-2.0
|
||||||
|
#
|
||||||
|
# Unless required by applicable law or agreed to in writing, software
|
||||||
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||||
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||||
|
# See the License for the specific language governing permissions and
|
||||||
|
# limitations under the License.
|
||||||
|
from typing import Dict
|
||||||
|
from typing import List
|
||||||
|
|
||||||
|
import numpy as np
|
||||||
|
import paddle
|
||||||
|
import ToJyutping
|
||||||
|
|
||||||
|
from paddlespeech.t2s.frontend.zh_normalization.text_normlization import TextNormalizer
|
||||||
|
|
||||||
|
INITIALS = [
|
||||||
|
'p', 'b', 't', 'd', 'ts', 'dz', 'k', 'g', 'kw', 'gw', 'f', 'h', 'l', 'm',
|
||||||
|
'ng', 'n', 's', 'y', 'w', 'c', 'z', 'j'
|
||||||
|
]
|
||||||
|
INITIALS += ['sp', 'spl', 'spn', 'sil']
|
||||||
|
|
||||||
|
|
||||||
|
def get_lines(cantons: List[str]):
|
||||||
|
phones = []
|
||||||
|
for canton in cantons:
|
||||||
|
for consonant in INITIALS:
|
||||||
|
if canton.startswith(consonant):
|
||||||
|
c, v = canton[:len(consonant)], canton[len(consonant):]
|
||||||
|
phones = phones + [c, v]
|
||||||
|
return phones
|
||||||
|
|
||||||
|
|
||||||
|
class CantonFrontend():
|
||||||
|
def __init__(self, phone_vocab_path: str):
|
||||||
|
self.text_normalizer = TextNormalizer()
|
||||||
|
self.punc = "：，；。？！“”‘’':,;.?!"
|
||||||
|
|
||||||
|
self.vocab_phones = {}
|
||||||
|
if phone_vocab_path:
|
||||||
|
with open(phone_vocab_path, 'rt', encoding='utf-8') as f:
|
||||||
|
phn_id = [line.strip().split() for line in f.readlines()]
|
||||||
|
for phn, id in phn_id:
|
||||||
|
self.vocab_phones[phn] = int(id)
|
||||||
|
|
||||||
|
# if merge_sentences, merge all sentences into one phone sequence
|
||||||
|
def _g2p(self, sentences: List[str],
|
||||||
|
merge_sentences: bool=True) -> List[List[str]]:
|
||||||
|
phones_list = []
|
||||||
|
for sentence in sentences:
|
||||||
|
phones_str = ToJyutping.get_jyutping_text(sentence)
|
||||||
|
phones_split = get_lines(phones_str.split(' '))
|
||||||
|
phones_list.append(phones_split)
|
||||||
|
return phones_list
|
||||||
|
|
||||||
|
def _p2id(self, phonemes: List[str]) -> np.ndarray:
|
||||||
|
# replace unk phone with sp
|
||||||
|
phonemes = [
|
||||||
|
phn if phn in self.vocab_phones else "sp" for phn in phonemes
|
||||||
|
]
|
||||||
|
phone_ids = [self.vocab_phones[item] for item in phonemes]
|
||||||
|
return np.array(phone_ids, np.int64)
|
||||||
|
|
||||||
|
def get_phonemes(self,
|
||||||
|
sentence: str,
|
||||||
|
merge_sentences: bool=True,
|
||||||
|
print_info: bool=False) -> List[List[str]]:
|
||||||
|
sentences = self.text_normalizer.normalize(sentence)
|
||||||
|
phonemes = self._g2p(sentences, merge_sentences=merge_sentences)
|
||||||
|
if print_info:
|
||||||
|
print("----------------------------")
|
||||||
|
print("text norm results:")
|
||||||
|
print(sentences)
|
||||||
|
print("----------------------------")
|
||||||
|
print("g2p results:")
|
||||||
|
print(phonemes)
|
||||||
|
print("----------------------------")
|
||||||
|
return phonemes
|
||||||
|
|
||||||
|
def get_input_ids(self,
|
||||||
|
sentence: str,
|
||||||
|
merge_sentences: bool=True,
|
||||||
|
print_info: bool=False,
|
||||||
|
to_tensor: bool=True) -> Dict[str, List[paddle.Tensor]]:
|
||||||
|
|
||||||
|
phonemes = self.get_phonemes(
|
||||||
|
sentence, merge_sentences=merge_sentences, print_info=print_info)
|
||||||
|
result = {}
|
||||||
|
temp_phone_ids = []
|
||||||
|
|
||||||
|
for phones in phonemes:
|
||||||
|
if phones:
|
||||||
|
phone_ids = self._p2id(phones)
|
||||||
|
# if use paddle.to_tensor() in onnxruntime, the first time will be too low
|
||||||
|
if to_tensor:
|
||||||
|
phone_ids = paddle.to_tensor(phone_ids)
|
||||||
|
temp_phone_ids.append(phone_ids)
|
||||||
|
if temp_phone_ids:
|
||||||
|
result["phone_ids"] = temp_phone_ids
|
||||||
|
return result
|
||||||
Loading…
Reference in new issue