Merge pull request #992 from yt605155624/fix_docs

[TTS] add tts tutorial

demos/metaverse/path.sh

+#!/bin/bash
+export MAIN_ROOT=`realpath ${PWD}/../../`
+
+export PATH=${MAIN_ROOT}:${MAIN_ROOT}/utils:${PATH}
+export LC_ALL=C
+
+export PYTHONDONTWRITEBYTECODE=1
+# Use UTF-8 in Python to avoid UnicodeDecodeError when LC_ALL=C
+export PYTHONIOENCODING=UTF-8
+export PYTHONPATH=${MAIN_ROOT}:${PYTHONPATH}
+
+MODEL=fastspeech2
+export BIN_DIR=${MAIN_ROOT}/paddlespeech/t2s/exps/${MODEL}

demos/metaverse/run.sh

+#!/bin/bash
+source path.sh
+
+gpus=0
+stage=0
+stop_stage=100
+
+# with the following command, you can choice the stage range you want to run
+# such as `./run.sh --stage 0 --stop-stage 0`
+# this can not be mixed use with `$1`, `$2` ...
+source ${MAIN_ROOT}/utils/parse_options.sh || exit 1
+
+mkdir -p download
+
+if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
+    # install PaddleGAN
+    git clone https://github.com/PaddlePaddle/PaddleGAN.git
+    pip install -e PaddleGAN/
+fi
+
+if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then 
+    # download pretrained PaddleGAN model
+    wget -P download https://paddlegan.bj.bcebos.com/models/wav2lip_hq.pdparams
+fi 
+
+if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
+    # download pretrained tts models and unzip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip
+    unzip -d download download/pwg_baker_ckpt_0.4.zip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip
+    unzip -d download download/fastspeech2_nosil_baker_ckpt_0.4.zip
+fi
+
+if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
+    # run tts
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    python3 ${BIN_DIR}/synthesize_e2e.py \
+        --fastspeech2-config=download/fastspeech2_nosil_baker_ckpt_0.4/default.yaml \
+        --fastspeech2-checkpoint=download/fastspeech2_nosil_baker_ckpt_0.4/snapshot_iter_76000.pdz \
+        --fastspeech2-stat=download/fastspeech2_nosil_baker_ckpt_0.4/speech_stats.npy \
+        --pwg-config=download/pwg_baker_ckpt_0.4/pwg_default.yaml \
+        --pwg-checkpoint=download/pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
+        --pwg-stat=download/pwg_baker_ckpt_0.4/pwg_stats.npy \
+        --text=sentences.txt \
+        --output-dir=output/wavs \
+        --inference-dir=output/inference \
+        --phones-dict=download/fastspeech2_nosil_baker_ckpt_0.4/phone_id_map.txt
+    # output/inference is not needed here, which save the static models
+    rm -rf output/inference
+fi
+
+if [ ${stage} -le  4 ] && [ ${stop_stage} -ge 4 ]; then
+    # We only test one audio here, cause it's slow
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    python3 PaddleGAN/applications/tools/wav2lip.py \
+        --checkpoint_path download/wav2lip_hq.pdparams \
+        --face Lamarr.png \
+        --audio output/wavs/000.wav \
+        --outfile output/tts_lips.mp4 \
+        --face_enhancement
+fi

demos/metaverse/sentences.txt

+000 谁知青蛙一落地，竟变成了一位英俊的王子。于是遵照国王的意思，他做了公主的亲密伴侣。

demos/story_talker/ocr.py

+# 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.
+import argparse
+import os
+import re
+from pathlib import Path
+
+import paddle
+from paddleocr import draw_ocr
+from paddleocr import PaddleOCR
+from PIL import Image
+
+
+def evaluate(args, ocr):
+    img_dir = Path(args.img_dir)
+    output_dir = Path(args.output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    img_out_dir = output_dir / "imgs"
+    img_out_dir.mkdir(parents=True, exist_ok=True)
+    with open(output_dir / "sentences.txt", "w") as wf:
+        for name in os.listdir(img_dir):
+            id = name.split(".")[0]
+            img_path = img_dir / name
+            result = ocr.ocr(str(img_path), cls=True)
+            # draw result
+            image = Image.open(img_path).convert('RGB')
+            boxes = [line[0] for line in result]
+            txts = [line[1][0] for line in result]
+            scores = [line[1][1] for line in result]
+            im_show = draw_ocr(
+                image, boxes, txts, scores, font_path=args.font_path)
+            im_show = Image.fromarray(im_show)
+            paragraph = "".join(txts)
+            # 过滤出中文结果
+            pattern = re.compile(r'[^(\u4e00-\u9fa5)+，。？、]')
+            sentence = re.sub(pattern, '', paragraph)
+            im_show.save(img_out_dir / name)
+            wf.write(id + " " + sentence + "\n")
+
+
+def main():
+    # parse args and config and redirect to train_sp
+    parser = argparse.ArgumentParser(
+        description="Synthesize with fastspeech2 & parallel wavegan.")
+    parser.add_argument("--img-dir", default="imgs", type=str, help="img_dir.")
+    parser.add_argument(
+        "--output-dir",
+        type=str,
+        default="output",
+        help="output sentences path.")
+    parser.add_argument(
+        "--font-path", type=str, default="simfang.ttf", help="font path")
+    args = parser.parse_args()
+
+    paddle.set_device("gpu")
+    # need to run only once to download and load model into memory
+    ocr = PaddleOCR(use_angle_cls=True, lang='ch')
+
+    evaluate(args, ocr)
+
+
+if __name__ == "__main__":
+    main()

demos/story_talker/path.sh

+#!/bin/bash
+export MAIN_ROOT=`realpath ${PWD}/../../`
+
+export PATH=${MAIN_ROOT}:${MAIN_ROOT}/utils:${PATH}
+export LC_ALL=C
+
+export PYTHONDONTWRITEBYTECODE=1
+# Use UTF-8 in Python to avoid UnicodeDecodeError when LC_ALL=C
+export PYTHONIOENCODING=UTF-8
+export PYTHONPATH=${MAIN_ROOT}:${PYTHONPATH}
+
+MODEL=fastspeech2
+export BIN_DIR=${MAIN_ROOT}/paddlespeech/t2s/exps/${MODEL}

demos/story_talker/run.sh

+#!/bin/bash
+source path.sh
+
+gpus=0
+stage=0
+stop_stage=100
+
+# with the following command, you can choice the stage range you want to run
+# such as `./run.sh --stage 0 --stop-stage 0`
+# this can not be mixed use with `$1`, `$2` ...
+source ${MAIN_ROOT}/utils/parse_options.sh || exit 1
+
+mkdir -p download
+
+if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
+    # install PaddleOCR
+    pip install "paddleocr>=2.0.1"
+fi
+
+if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
+    # download pretrained tts models and unzip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip
+    unzip -d download download/pwg_baker_ckpt_0.4.zip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip
+    unzip -d download download/fastspeech2_nosil_baker_ckpt_0.4.zip
+fi
+
+if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
+    # run ocr
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    python3 ocr.py --img-dir=imgs --output-dir=output --font-path=simfang.ttf
+fi
+
+if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
+    # run tts
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    python3 ${BIN_DIR}/synthesize_e2e.py \
+        --fastspeech2-config=download/fastspeech2_nosil_baker_ckpt_0.4/default.yaml \
+        --fastspeech2-checkpoint=download/fastspeech2_nosil_baker_ckpt_0.4/snapshot_iter_76000.pdz \
+        --fastspeech2-stat=download/fastspeech2_nosil_baker_ckpt_0.4/speech_stats.npy \
+        --pwg-config=download/pwg_baker_ckpt_0.4/pwg_default.yaml \
+        --pwg-checkpoint=download/pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
+        --pwg-stat=download/pwg_baker_ckpt_0.4/pwg_stats.npy \
+        --text=output/sentences.txt \
+        --output-dir=output/wavs \
+        --inference-dir=output/inference \
+        --phones-dict=download/fastspeech2_nosil_baker_ckpt_0.4/phone_id_map.txt
+    # output/inference is not needed here, which save the static models
+    rm -rf output/inference
+fi

demos/style_fs2/path.sh

+#!/bin/bash
+export MAIN_ROOT=`realpath ${PWD}/../../`
+
+export PATH=${MAIN_ROOT}:${MAIN_ROOT}/utils:${PATH}
+export LC_ALL=C
+
+export PYTHONDONTWRITEBYTECODE=1
+# Use UTF-8 in Python to avoid UnicodeDecodeError when LC_ALL=C
+export PYTHONIOENCODING=UTF-8
+export PYTHONPATH=${MAIN_ROOT}:${PYTHONPATH}
+MODEL=fastspeech2
+export BIN_DIR=${MAIN_ROOT}/paddlespeech/t2s/exps/${MODEL}

demos/style_fs2/run.sh

+#!/bin/bash
+source path.sh
+
+gpus=0
+stage=0
+stop_stage=100
+
+# with the following command, you can choice the stage range you want to run
+# such as `./run.sh --stage 0 --stop-stage 0`
+# this can not be mixed use with `$1`, `$2` ...
+source ${MAIN_ROOT}/utils/parse_options.sh || exit 1
+
+mkdir -p download
+
+if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
+    # download pretrained tts models and unzip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip
+    unzip -d download download/pwg_baker_ckpt_0.4.zip
+    wget -P download https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip
+    unzip -d download download/fastspeech2_nosil_baker_ckpt_0.4.zip
+fi
+
+if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
+    # run tts
+    CUDA_VISIBLE_DEVICES=${gpus} \
+    python3 style_syn.py \
+        --fastspeech2-config=download/fastspeech2_nosil_baker_ckpt_0.4/default.yaml \
+        --fastspeech2-checkpoint=download/fastspeech2_nosil_baker_ckpt_0.4/snapshot_iter_76000.pdz \
+        --fastspeech2-stat=download/fastspeech2_nosil_baker_ckpt_0.4/speech_stats.npy \
+        --fastspeech2-pitch-stat=download/fastspeech2_nosil_baker_ckpt_0.4/pitch_stats.npy \
+        --fastspeech2-energy-stat=download/fastspeech2_nosil_baker_ckpt_0.4/energy_stats.npy \
+        --pwg-config=download/pwg_baker_ckpt_0.4/pwg_default.yaml \
+        --pwg-checkpoint=download/pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
+        --pwg-stat=download/pwg_baker_ckpt_0.4/pwg_stats.npy \
+        --text=${BIN_DIR}/../sentences.txt \
+        --output-dir=output \
+        --phones-dict=download/fastspeech2_nosil_baker_ckpt_0.4/phone_id_map.txt
+fi

demos/style_fs2/sentences.txt

+000 谁知青蛙一落地，竟变成了一位英俊的王子。于是遵照国王的意思，他做了公主的亲密伴侣。

demos/style_fs2/style_syn.py

+# 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.
+import argparse
+from pathlib import Path
+from typing import Union
+
+import numpy as np
+import paddle
+import soundfile as sf
+import yaml
+from yacs.config import CfgNode
+
+from paddlespeech.t2s.frontend.zh_frontend import Frontend
+from paddlespeech.t2s.models.fastspeech2 import FastSpeech2
+from paddlespeech.t2s.models.fastspeech2 import FastSpeech2Inference
+from paddlespeech.t2s.models.parallel_wavegan import PWGGenerator
+from paddlespeech.t2s.models.parallel_wavegan import PWGInference
+from paddlespeech.t2s.modules.normalizer import ZScore
+
+
+class StyleFastSpeech2Inference(FastSpeech2Inference):
+    def __init__(self, normalizer, model, pitch_stats_path, energy_stats_path):
+        super().__init__(normalizer, model)
+        pitch_mean, pitch_std = np.load(pitch_stats_path)
+        self.pitch_mean = paddle.to_tensor(pitch_mean)
+        self.pitch_std = paddle.to_tensor(pitch_std)
+        energy_mean, energy_std = np.load(energy_stats_path)
+        self.energy_mean = paddle.to_tensor(energy_mean)
+        self.energy_std = paddle.to_tensor(energy_std)
+
+    def denorm(self, data, mean, std):
+        return data * std + mean
+
+    def norm(self, data, mean, std):
+        return (data - mean) / std
+
+    def forward(self,
+                text: paddle.Tensor,
+                durations: Union[paddle.Tensor, np.ndarray]=None,
+                durations_scale: Union[int, float]=None,
+                durations_bias: Union[int, float]=None,
+                pitch: Union[paddle.Tensor, np.ndarray]=None,
+                pitch_scale: Union[int, float]=None,
+                pitch_bias: Union[int, float]=None,
+                energy: Union[paddle.Tensor, np.ndarray]=None,
+                energy_scale: Union[int, float]=None,
+                energy_bias: Union[int, float]=None,
+                robot: bool=False):
+        """
+        Parameters
+        ----------
+        text : Tensor(int64)
+            Input sequence of characters (T,).
+        speech : Tensor, optional
+            Feature sequence to extract style (N, idim).
+        durations : paddle.Tensor/np.ndarray, optional (int64)
+            Groundtruth of duration (T,), this will overwrite the set of durations_scale and durations_bias
+        durations_scale: int/float, optional
+        durations_bias: int/float, optional
+        pitch : paddle.Tensor/np.ndarray, optional
+            Groundtruth of token-averaged pitch (T, 1), this will overwrite the set of pitch_scale and pitch_bias
+        pitch_scale: int/float, optional
+            In denormed HZ domain.
+        pitch_bias: int/float, optional
+            In denormed HZ domain.
+        energy : paddle.Tensor/np.ndarray, optional
+            Groundtruth of token-averaged energy (T, 1), this will overwrite the set of energy_scale and energy_bias
+        energy_scale: int/float, optional
+            In denormed domain.
+        energy_bias: int/float, optional
+            In denormed domain.
+        robot : bool, optional
+            Weather output robot style
+        Returns
+        ----------
+        Tensor
+            Output sequence of features (L, odim).
+        """
+        normalized_mel, d_outs, p_outs, e_outs = self.acoustic_model.inference(
+            text, durations=None, pitch=None, energy=None)
+        # priority: groundtruth > scale/bias > previous output
+        # set durations
+        if isinstance(durations, np.ndarray):
+            durations = paddle.to_tensor(durations)
+        elif isinstance(durations, paddle.Tensor):
+            durations = durations
+        elif durations_scale or durations_bias:
+            durations_scale = durations_scale if durations_scale is not None else 1
+            durations_bias = durations_bias if durations_bias is not None else 0
+            durations = durations_scale * d_outs + durations_bias
+        else:
+            durations = d_outs
+
+        if robot:
+            # set normed pitch to zeros have the same effect with set denormd ones to mean
+            pitch = paddle.zeros(p_outs.shape)
+
+        # set pitch, can overwrite robot set  
+        if isinstance(pitch, np.ndarray):
+            pitch = paddle.to_tensor(pitch)
+        elif isinstance(pitch, paddle.Tensor):
+            pitch = pitch
+        elif pitch_scale or pitch_bias:
+            pitch_scale = pitch_scale if pitch_scale is not None else 1
+            pitch_bias = pitch_bias if pitch_bias is not None else 0
+            p_Hz = paddle.exp(
+                self.denorm(p_outs, self.pitch_mean, self.pitch_std))
+            p_HZ = pitch_scale * p_Hz + pitch_bias
+            pitch = self.norm(paddle.log(p_HZ), self.pitch_mean, self.pitch_std)
+        else:
+            pitch = p_outs
+
+        # set energy
+        if isinstance(energy, np.ndarray):
+            energy = paddle.to_tensor(energy)
+        elif isinstance(energy, paddle.Tensor):
+            energy = energy
+        elif energy_scale or energy_bias:
+            energy_scale = energy_scale if energy_scale is not None else 1
+            energy_bias = energy_bias if energy_bias is not None else 0
+            e_dnorm = self.denorm(e_outs, self.energy_mean, self.energy_std)
+            e_dnorm = energy_scale * e_dnorm + energy_bias
+            energy = self.norm(e_dnorm, self.energy_mean, self.energy_std)
+        else:
+            energy = e_outs
+
+        normalized_mel, d_outs, p_outs, e_outs = self.acoustic_model.inference(
+            text,
+            durations=durations,
+            pitch=pitch,
+            energy=energy,
+            use_teacher_forcing=True)
+
+        logmel = self.normalizer.inverse(normalized_mel)
+        return logmel
+
+
+def evaluate(args, fastspeech2_config, pwg_config):
+
+    # construct dataset for evaluation
+    sentences = []
+    with open(args.text, 'rt') as f:
+        for line in f:
+            utt_id, sentence = line.strip().split()
+            sentences.append((utt_id, sentence))
+
+    with open(args.phones_dict, "r") as f:
+        phn_id = [line.strip().split() for line in f.readlines()]
+    vocab_size = len(phn_id)
+    print("vocab_size:", vocab_size)
+
+    odim = fastspeech2_config.n_mels
+    model = FastSpeech2(
+        idim=vocab_size, odim=odim, **fastspeech2_config["model"])
+
+    model.set_state_dict(
+        paddle.load(args.fastspeech2_checkpoint)["main_params"])
+    model.eval()
+
+    vocoder = PWGGenerator(**pwg_config["generator_params"])
+    vocoder.set_state_dict(paddle.load(args.pwg_checkpoint)["generator_params"])
+    vocoder.remove_weight_norm()
+    vocoder.eval()
+    print("model done!")
+
+    frontend = Frontend(phone_vocab_path=args.phones_dict)
+    print("frontend done!")
+
+    stat = np.load(args.fastspeech2_stat)
+    mu, std = stat
+    mu = paddle.to_tensor(mu)
+    std = paddle.to_tensor(std)
+    fastspeech2_normalizer = ZScore(mu, std)
+
+    stat = np.load(args.pwg_stat)
+    mu, std = stat
+    mu = paddle.to_tensor(mu)
+    std = paddle.to_tensor(std)
+    pwg_normalizer = ZScore(mu, std)
+
+    fastspeech2_inference = StyleFastSpeech2Inference(
+        fastspeech2_normalizer, model, args.fastspeech2_pitch_stat,
+        args.fastspeech2_energy_stat)
+    fastspeech2_inference.eval()
+
+    pwg_inference = PWGInference(pwg_normalizer, vocoder)
+    pwg_inference.eval()
+
+    output_dir = Path(args.output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    styles = ["normal", "robot", "1.2xspeed", "0.8xspeed", "child_voice"]
+    for style in styles:
+        robot = False
+        durations = None
+        durations_scale = None
+        durations_bias = None
+        pitch = None
+        pitch_scale = None
+        pitch_bias = None
+        energy = None
+        energy_scale = None
+        energy_bias = None
+        if style == "robot":
+            # all tones in phones be `1`
+            # all pitch should be the same, we use mean here
+            robot = True
+        if style == "1.2xspeed":
+            durations_scale = 1 / 1.2
+        if style == "0.8xspeed":
+            durations_scale = 1 / 0.8
+        if style == "child_voice":
+            pitch_scale = 1.3
+        sub_output_dir = output_dir / style
+        sub_output_dir.mkdir(parents=True, exist_ok=True)
+        for utt_id, sentence in sentences:
+            input_ids = frontend.get_input_ids(
+                sentence, merge_sentences=True, robot=robot)
+            phone_ids = input_ids["phone_ids"][0]
+
+            with paddle.no_grad():
+                mel = fastspeech2_inference(
+                    phone_ids,
+                    durations=durations,
+                    durations_scale=durations_scale,
+                    durations_bias=durations_bias,
+                    pitch=pitch,
+                    pitch_scale=pitch_scale,
+                    pitch_bias=pitch_bias,
+                    energy=energy,
+                    energy_scale=energy_scale,
+                    energy_bias=energy_bias,
+                    robot=robot)
+                wav = pwg_inference(mel)
+
+            sf.write(
+                str(sub_output_dir / (utt_id + ".wav")),
+                wav.numpy(),
+                samplerate=fastspeech2_config.fs)
+            print(f"{style}_{utt_id} done!")
+
+
+def main():
+    # parse args and config and redirect to train_sp
+    parser = argparse.ArgumentParser(
+        description="Synthesize with fastspeech2 & parallel wavegan.")
+    parser.add_argument(
+        "--fastspeech2-config", type=str, help="fastspeech2 config file.")
+    parser.add_argument(
+        "--fastspeech2-checkpoint",
+        type=str,
+        help="fastspeech2 checkpoint to load.")
+    parser.add_argument(
+        "--fastspeech2-stat",
+        type=str,
+        help="mean and standard deviation used to normalize spectrogram when training fastspeech2."
+    )
+    parser.add_argument(
+        "--fastspeech2-pitch-stat",
+        type=str,
+        help="mean and standard deviation used to normalize pitch when training fastspeech2"
+    )
+    parser.add_argument(
+        "--fastspeech2-energy-stat",
+        type=str,
+        help="mean and standard deviation used to normalize energy when training fastspeech2."
+    )
+    parser.add_argument(
+        "--pwg-config", type=str, help="parallel wavegan config file.")
+    parser.add_argument(
+        "--pwg-checkpoint",
+        type=str,
+        help="parallel wavegan generator parameters to load.")
+    parser.add_argument(
+        "--pwg-stat",
+        type=str,
+        help="mean and standard deviation used to normalize spectrogram when training parallel wavegan."
+    )
+    parser.add_argument(
+        "--phones-dict",
+        type=str,
+        default="phone_id_map.txt",
+        help="phone vocabulary file.")
+    parser.add_argument(
+        "--text",
+        type=str,
+        help="text to synthesize, a 'utt_id sentence' pair per line.")
+    parser.add_argument("--output-dir", type=str, help="output dir.")
+    parser.add_argument(
+        "--ngpu", type=int, default=1, help="if ngpu == 0, use cpu.")
+    parser.add_argument("--verbose", type=int, default=1, help="verbose.")
+
+    args = parser.parse_args()
+
+    if args.ngpu == 0:
+        paddle.set_device("cpu")
+    elif args.ngpu > 0:
+        paddle.set_device("gpu")
+    else:
+        print("ngpu should >= 0 !")
+
+    with open(args.fastspeech2_config) as f:
+        fastspeech2_config = CfgNode(yaml.safe_load(f))
+    with open(args.pwg_config) as f:
+        pwg_config = CfgNode(yaml.safe_load(f))
+
+    print("========Args========")
+    print(yaml.safe_dump(vars(args)))
+    print("========Config========")
+    print(fastspeech2_config)
+    print(pwg_config)
+
+    evaluate(args, fastspeech2_config, pwg_config)
+
+
+if __name__ == "__main__":
+    main()

docs/source/asr/models_introduction.md

@@ -13,7 +13,7 @@ In addition, the training process and the testing process are also introduced.
 The arcitecture of the model is shown in Fig.1.
 
 <p align="center">
-    <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/ds2onlineModel.png" width=800>
+    <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/ds2onlineModel.png" width=800>
     <br/>Fig.1 The Arcitecture of deepspeech2 online model
 </p>
 
@@ -160,7 +160,7 @@ The deepspeech2 offline model is similarity to the deepspeech2 online model. The
 
 The arcitecture of the model is shown in Fig.2.
 <p align="center">
-    <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/ds2offlineModel.png" width=800>
+    <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/ds2offlineModel.png" width=800>
     <br/>Fig.2 The Arcitecture of deepspeech2 offline model
 </p>
 

docs/source/released_model.md

+
 # Released Models
 
 ## Speech-To-Text Models
@@ -28,28 +29,29 @@ Language Model | Training Data | Token-based | Size | Descriptions
 
 ## Text-To-Speech Models
 ### Acoustic Models
-Model Type | Dataset| Example Link | Pretrained Models
-:-------------:| :------------:| :-----: | :-----
-Tacotron2|LJSpeech|[tacotron2-vctk](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts0)|[tacotron2_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/tacotron2_ljspeech_ckpt_0.3.zip)
-TransformerTTS| LJSpeech| [transformer-ljspeech](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts1)|[transformer_tts_ljspeech_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/transformer_tts_ljspeech_ckpt_0.4.zip)
-SpeedySpeech| CSMSC | [speedyspeech-csmsc](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts2) |[speedyspeech_nosil_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_ckpt_0.5.zip)
-FastSpeech2| CSMSC |[fastspeech2-csmsc](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts3)|[fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip)
-FastSpeech2| AISHELL-3 |[fastspeech2-aishell3](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/aishell3/tts3)|[fastspeech2_nosil_aishell3_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_aishell3_ckpt_0.4.zip)
-FastSpeech2| LJSpeech |[fastspeech2-ljspeech](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts3)|[fastspeech2_nosil_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_ljspeech_ckpt_0.5.zip)
-FastSpeech2| VCTK |[fastspeech2-csmsc](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/vctk/tts3)|[fastspeech2_nosil_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_vctk_ckpt_0.5.zip)
-
+Model Type | Dataset| Example Link | Pretrained Models|Static Models|Siize(static)
+:-------------:| :------------:| :-----: | :-----:| :-----:| :-----:
+Tacotron2|LJSpeech|[tacotron2-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts0)|[tacotron2_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/tacotron2_ljspeech_ckpt_0.3.zip)|||
+TransformerTTS| LJSpeech| [transformer-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts1)|[transformer_tts_ljspeech_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/transformer_tts_ljspeech_ckpt_0.4.zip)|||
+SpeedySpeech| CSMSC | [speedyspeech-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts2) |[speedyspeech_nosil_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_ckpt_0.5.zip)|[speedyspeech_nosil_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_static_0.5.zip)|12MB|
+FastSpeech2| CSMSC |[fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3)|[fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip)|[fastspeech2_nosil_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_static_0.4.zip)|157MB|
+FastSpeech2| AISHELL-3 |[fastspeech2-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/tts3)|[fastspeech2_nosil_aishell3_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_aishell3_ckpt_0.4.zip)|||
+FastSpeech2| LJSpeech |[fastspeech2-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts3)|[fastspeech2_nosil_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_ljspeech_ckpt_0.5.zip)|||
+FastSpeech2| VCTK |[fastspeech2-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/tts3)|[fastspeech2_nosil_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_vctk_ckpt_0.5.zip)|||
 
 ### Vocoders
 
-Model Type | Dataset| Example Link | Pretrained Models
-:-------------:| :------------:| :-----: | :-----
-WaveFlow| LJSpeech |[waveflow-ljspeech](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0)|[waveflow_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_ljspeech_ckpt_0.3.zip)
-Parallel WaveGAN| CSMSC |[PWGAN-csmsc](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1)|[pwg_baker_ckpt_0.4.zip.](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip)
-Parallel WaveGAN| LJSpeech |[PWGAN-ljspeech](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc1)|[pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_ljspeech_ckpt_0.5.zip)
-Parallel WaveGAN| VCTK |[PWGAN-vctk](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/vctk/voc1)|[pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_vctk_ckpt_0.5.zip)
+Model Type | Dataset| Example Link | Pretrained Models| Static Models|Size(static)
+:-------------:| :------------:| :-----: | :-----:| :-----:| :-----:
+WaveFlow| LJSpeech |[waveflow-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0)|[waveflow_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_ljspeech_ckpt_0.3.zip)|||
+Parallel WaveGAN| CSMSC |[PWGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1)|[pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip)|[pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_static_0.4.zip)|5.1MB|
+Parallel WaveGAN| LJSpeech |[PWGAN-ljspeech](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc1)|[pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_ljspeech_ckpt_0.5.zip)|||
+Parallel WaveGAN|AISHELL-3 |[PWGAN-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1)|[pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_aishell3_ckpt_0.5.zip)|||
+Parallel WaveGAN| VCTK |[PWGAN-vctk](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc1)|[pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_vctk_ckpt_0.5.zip)|||
+|Multi Band MelGAN |CSMSC|[MB MelGAN-csmsc](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc3) | [mb_melgan_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_ckpt_0.5.zip)|[mb_melgan_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_static_0.5.zip) |8.2MB|
 
 ### Voice Cloning
 Model Type | Dataset| Example Link | Pretrained Models
-:-------------:| :------------:| :-----: | :-----
-GE2E| AISHELL-3, etc. |[ge2e](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/ge2e)|[ge2e_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/ge2e_ckpt_0.3.zip)
-GE2E + Tactron2| AISHELL-3 |[ge2e-tactron2-aishell3](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/aishell3/vc0)|[tacotron2_aishell3_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/tacotron2_aishell3_ckpt_0.3.zip)
+:-------------:| :------------:| :-----: | :-----:
+GE2E| AISHELL-3, etc. |[ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e)|[ge2e_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/ge2e_ckpt_0.3.zip)
+GE2E + Tactron2| AISHELL-3 |[ge2e-tactron2-aishell3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/vc0)|[tacotron2_aishell3_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/tacotron2_aishell3_ckpt_0.3.zip)

docs/source/tts/gan_vocoder.md

@@ -6,4 +6,4 @@ Model  | Generator Loss |Discriminator Loss
 Parallel Wave GAN| adversial loss <br> Feature Matching  | Multi-Scale Discriminator |
 Mel GAN |adversial loss <br> Multi-resolution STFT loss  | adversial loss|
 Multi-Band Mel GAN | adversial loss <br> full band Multi-resolution STFT loss <br> sub band Multi-resolution STFT loss |Multi-Scale Discriminator|
-HiFi GAN |adversial loss <br> Feature Matching <br>  Mel-Spectrogram Loss | Multi-Scale Discriminator <br> Multi-Period Discriminato  |
+HiFi GAN |adversial loss <br> Feature Matching <br>  Mel-Spectrogram Loss | Multi-Scale Discriminator <br> Multi-Period Discriminator|

docs/source/tts/models_introduction.md

@@ -27,14 +27,14 @@ At present, there are two mainstream acoustic model structures.
    - Acoustic decoder (N Frames - > N Frames).
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/frame_level_am.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/frame_level_am.png" width=500 /> <br>
 </div>
 
 - Sequence to sequence acoustic model:
     - M Tokens - > N Frames.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/seq2seq_am.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/seq2seq_am.png" width=500 /> <br>
 </div>
 
 ### Tacotron2
@@ -54,7 +54,7 @@ At present, there are two mainstream acoustic model structures.
     - CBHG postprocess.
     - Vocoder: Griffin-Lim.
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/tacotron.png" width=700 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/tacotron.png" width=700 /> <br>
 </div>
 
 **Advantage of Tacotron:**
@@ -89,10 +89,10 @@ At present, there are two mainstream acoustic model structures.
    - The alignment matrix of previous time is considered at the step `t` of decoder.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/tacotron2.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/tacotron2.png" width=500 /> <br>
 </div>
 
-You can find PaddleSpeech TTS's tacotron2 with LJSpeech dataset example at [examples/ljspeech/tts0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts0).
+You can find PaddleSpeech TTS's tacotron2 with LJSpeech dataset example at [examples/ljspeech/tts0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts0).
 
 ### TransformerTTS
 **Disadvantages of the Tacotrons:**
@@ -118,7 +118,7 @@ Transformer TTS is a combination of Tacotron2 and Transformer.
     - Positional Encoding.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/transformer.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/transformer.png" width=500 /> <br>
 </div>
 
 #### Transformer TTS
@@ -138,7 +138,7 @@ Transformer TTS is a seq2seq acoustic model based on Transformer and Tacotron2.
     - Uniform scale position encoding may have a negative impact on input or output sequences.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/transformer_tts.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/transformer_tts.png" width=500 /> <br>
 </div>
 
 **Disadvantages of Transformer TTS:**
@@ -146,7 +146,7 @@ Transformer TTS is a seq2seq acoustic model based on Transformer and Tacotron2.
 - The ability to perceive local information is weak, and local information is more related to pronunciation.
 - Stability is worse than Tacotron2.
 
-You can find PaddleSpeech TTS's Transformer TTS with LJSpeech dataset example at [examples/ljspeech/tts1](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/tts1).
+You can find PaddleSpeech TTS's Transformer TTS with LJSpeech dataset example at [examples/ljspeech/tts1](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/tts1).
 
 
 ### FastSpeech2
@@ -184,14 +184,14 @@ Instead of using the encoder-attention-decoder based architecture as adopted by
 • Can be generated in parallel (decoding time is less affected by sequence length)
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/fastspeech.png" width=800 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/fastspeech.png" width=800 /> <br>
 </div>
 
 #### FastPitch
 [FastPitch](https://arxiv.org/abs/2006.06873) follows FastSpeech. A single pitch value is predicted for every temporal location, which improves the overall quality of synthesized speech.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/fastpitch.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/fastpitch.png" width=500 /> <br>
 </div>
 
 #### FastSpeech2
@@ -209,10 +209,10 @@ Instead of using the encoder-attention-decoder based architecture as adopted by
 FastSpeech2 is similar to FastPitch but introduces more variation information of speech.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/fastspeech2.png" width=800 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/fastspeech2.png" width=800 /> <br>
 </div>
 
-You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example at [examples/csmsc/tts3](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts3), We use token-averaged pitch and energy values introduced in FastPitch rather than frame level ones in FastSpeech2.
+You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example at [examples/csmsc/tts3](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts3), We use token-averaged pitch and energy values introduced in FastPitch rather than frame level ones in FastSpeech2.
 
 ### SpeedySpeech
 [SpeedySpeech](https://arxiv.org/abs/2008.03802) simplify the teacher-student architecture of FastSpeech and provide a fast and stable training procedure.
@@ -223,10 +223,10 @@ You can find PaddleSpeech TTS's FastSpeech2/FastPitch with CSMSC dataset example
 - Describe a simple data augmentation technique that can be used early in the training to make the teacher network robust to sequential error propagation.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/speedyspeech.png" width=500 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/speedyspeech.png" width=500 /> <br>
 </div>
 
-You can find PaddleSpeech TTS's SpeedySpeech with CSMSC dataset example at [examples/csmsc/tts2](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/tts2).
+You can find PaddleSpeech TTS's SpeedySpeech with CSMSC dataset example at [examples/csmsc/tts2](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/tts2).
 
 ## Vocoders
 In speech synthesis, the main task of the vocoder is to convert the spectral parameters predicted by the acoustic model into the final speech waveform.
@@ -276,7 +276,7 @@ Here, we introduce a Flow-based vocoder WaveFlow and a GAN-based vocoder Paralle
 - It is a small-footprint flow-based model for raw audio. It has only 5.9M parameters, which is 15x smalller than WaveGlow (87.9M).
 - It is directly trained with maximum likelihood without probability density distillation and auxiliary losses as used in [Parallel WaveNet](https://arxiv.org/abs/1711.10433) and [ClariNet](https://openreview.net/pdf?id=HklY120cYm), which simplifies the training pipeline and reduces the cost of development.
 
-You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examples/ljspeech/voc0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0).
+You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examples/ljspeech/voc0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
 
 ### Parallel WaveGAN
 [Parallel WaveGAN](https://arxiv.org/abs/1910.11480) trains a non-autoregressive WaveNet variant as a generator in a GAN based training method.
@@ -289,7 +289,7 @@ You can find PaddleSpeech TTS's WaveFlow with LJSpeech dataset example at [examp
 - Multi-resolution STFT loss.
 
 <div align="left">
-  <img src="https://raw.githubusercontent.com/PaddlePaddle/DeepSpeech/develop/docs/images/pwg.png" width=600 /> <br>
+  <img src="https://raw.githubusercontent.com/PaddlePaddle/PaddleSpeech/develop/docs/images/pwg.png" width=600 /> <br>
 </div>
 
-You can find PaddleSpeech TTS's Parallel WaveGAN with CSMSC example at [examples/csmsc/voc1](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1).
+You can find PaddleSpeech TTS's Parallel WaveGAN with CSMSC example at [examples/csmsc/voc1](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1).

docs/source/tts/quick_start.md

@@ -18,7 +18,7 @@ The models in PaddleSpeech TTS have the following mapping relationship:
 
 ## Quick Start
 
-Let's take a FastSpeech2 + Parallel WaveGAN with CSMSC dataset for instance. (./examples/csmsc/)(https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc)
+Let's take a FastSpeech2 + Parallel WaveGAN with CSMSC dataset for instance. (./examples/csmsc/)(https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc)
 
 ### Train Parallel WaveGAN with CSMSC
 - Go to directory

docs/source/tts/zh_text_frontend.md

 # Chinese Rule Based Text Frontend
-A TTS system mainly includes three modules: `Text Frontend`, `Acoustic model` and `Vocoder`. We provide a complete Chinese text frontend module in PaddleSpeech TTS, see exapmle in [examples/other/text_frontend/](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/text_frontend).
+A TTS system mainly includes three modules: `Text Frontend`, `Acoustic model` and `Vocoder`. We provide a complete Chinese text frontend module in PaddleSpeech TTS, see exapmle in [examples/other/text_frontend/](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/text_frontend).
 
 A text frontend module mainly includes:
  - Text Segmentation

examples/aishell3/tts3/README.md

@@ -17,7 +17,7 @@ tar zxvf data_aishell3.tgz -C data_aishell3
 ```
 ### Get MFA result of AISHELL-3 and Extract it
 We use [MFA2.x](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for aishell3_fastspeech2.
-You can download from here [aishell3_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/AISHELL-3/with_tone/aishell3_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
+You can download from here [aishell3_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/AISHELL-3/with_tone/aishell3_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/data_aishell3`.
@@ -96,17 +96,17 @@ optional arguments:
 6. `--speaker-dict`is the path of the  speaker id map file when training a multi-speaker FastSpeech2.
 
 ### Synthesize
-We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
-Download pretrained parallel wavegan model from [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip) and unzip it.
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1) as the neural vocoder.
+Download pretrained parallel wavegan model from [pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_aishell3_ckpt_0.5.zip) and unzip it.
 ```bash
-unzip pwg_baker_ckpt_0.4.zip
+unzip pwg_aishell3_ckpt_0.5.zip
 ```
 Parallel WaveGAN checkpoint contains files listed below.
 ```text
-pwg_baker_ckpt_0.4
-├── pwg_default.yaml               # default config used to train parallel wavegan
-├── pwg_snapshot_iter_400000.pdz   # model parameters of parallel wavegan
-└── pwg_stats.npy                  # statistics used to normalize spectrogram when training parallel wavegan
+pwg_aishell3_ckpt_0.5
+├── default.yaml                   # default config used to train parallel wavegan
+├── feats_stats.npy                # statistics used to normalize spectrogram when training parallel wavegan
+└── snapshot_iter_1000000.pdz      # generator parameters of parallel wavegan
 ```
 `./local/synthesize.sh` calls `${BIN_DIR}/synthesize.py`, which can synthesize waveform from `metadata.jsonl`.
 ```bash
@@ -224,14 +224,12 @@ python3 ${BIN_DIR}/multi_spk_synthesize_e2e.py \
   --fastspeech2-config=fastspeech2_nosil_aishell3_ckpt_0.4/default.yaml \
   --fastspeech2-checkpoint=fastspeech2_nosil_aishell3_ckpt_0.4/snapshot_iter_96400.pdz \
   --fastspeech2-stat=fastspeech2_nosil_aishell3_ckpt_0.4/speech_stats.npy \
-  --pwg-config=pwg_baker_ckpt_0.4/pwg_default.yaml \
-  --pwg-checkpoint=pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
-  --pwg-stat=pwg_baker_ckpt_0.4/pwg_stats.npy \
+  --pwg-config=pwg_aishell3_ckpt_0.5/default.yaml \
+  --pwg-checkpoint=pwg_aishell3_ckpt_0.5/snapshot_iter_1000000.pdz \
+  --pwg-stat=pwg_aishell3_ckpt_0.5/feats_stats.npy  \
   --text=${BIN_DIR}/../sentences.txt \
   --output-dir=exp/default/test_e2e \
   --phones-dict=fastspeech2_nosil_aishell3_ckpt_0.4/phone_id_map.txt \
   --speaker-dict=fastspeech2_nosil_aishell3_ckpt_0.4/speaker_id_map.txt
 
 ```
-## Future work
-A multi-speaker  vocoder is needed.

examples/aishell3/vc0/README.md

 # Tacotron2 + AISHELL-3 Voice Cloning
 This example contains code used to train a [Tacotron2 ](https://arxiv.org/abs/1712.05884) model with [AISHELL-3](http://www.aishelltech.com/aishell_3). The trained model can be used in Voice Cloning Task, We refer to the model structure of  [Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis](https://arxiv.org/pdf/1806.04558.pdf) . The general steps are as follows:
-1. Speaker Encoder: We  use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the  transcriptions are not needed, we use more datasets, refer to  [ge2e](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/ge2e).
+1. Speaker Encoder: We  use a Speaker Verification to train a speaker encoder. Datasets used in this task are different from those used in Tacotron2, because the  transcriptions are not needed, we use more datasets, refer to  [ge2e](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/ge2e).
 2. Synthesizer: Then, we use the trained speaker encoder to generate utterance embedding for each  sentence in AISHELL-3. This embedding is a extra input of  Tacotron2 which will be concated with encoder outputs.
-3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0).
+3. Vocoder: We use WaveFlow as the neural Vocoder, refer to [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0).
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/data_aishell3`.
@@ -39,9 +39,9 @@ There are silence in the edge of AISHELL-3's wavs, and the audio amplitude is ve
 
 We use Montreal Force Aligner 1.0. The label in  aishell3 include pinyin，so the lexicon we provided to MFA is pinyin rather than Chinese characters. And the prosody marks(`$`  and `%`) need to be removed. You shoud preprocess the dataset into the format  which MFA needs, the texts have the same name with wavs and have the suffix `.lab`.
 
-We use [lexicon.txt](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/paddlespeech/t2s/exps/voice_cloning/tacotron2_ge2e/lexicon.txt) as the lexicon.
+We use [lexicon.txt](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/paddlespeech/t2s/exps/voice_cloning/tacotron2_ge2e/lexicon.txt) as the lexicon.
 
-You can download the alignment results from here [alignment_aishell3.tar.gz](https://paddlespeech.bj.bcebos.com/Parakeet/alignment_aishell3.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
+You can download the alignment results from here [alignment_aishell3.tar.gz](https://paddlespeech.bj.bcebos.com/Parakeet/alignment_aishell3.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
 
 ```bash
 if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then

examples/aishell3/voc1/README.md

+# Parallel WaveGAN with AISHELL-3
+This example contains code used to train a [parallel wavegan](http://arxiv.org/abs/1910.11480) model with [AISHELL-3](http://www.aishelltech.com/aishell_3).
+
+AISHELL-3 is a large-scale and high-fidelity multi-speaker Mandarin speech corpus which could be used to train multi-speaker Text-to-Speech (TTS) systems.
+## Dataset
+### Download and Extract the datasaet
+Download AISHELL-3.
+```bash
+wget https://www.openslr.org/resources/93/data_aishell3.tgz
+```
+Extract AISHELL-3.
+```bash
+mkdir data_aishell3
+tar zxvf data_aishell3.tgz -C data_aishell3
+```
+### Get MFA result of AISHELL-3 and Extract it
+We use [MFA2.x](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for aishell3_fastspeech2.
+You can download from here [aishell3_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/AISHELL-3/with_tone/aishell3_alignment_tone.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) (use MFA1.x now) of our repo.
+
+## Get Started
+Assume the path to the dataset is `~/datasets/data_aishell3`.
+Assume the path to the MFA result of AISHELL-3 is `./aishell3_alignment_tone`.
+Run the command below to
+1. **source path**.
+2. preprocess the dataset,
+3. train the model.
+4. synthesize wavs.
+    - synthesize waveform from `metadata.jsonl`.
+```bash
+./run.sh
+```
+### Preprocess the dataset
+```bash
+./local/preprocess.sh ${conf_path}
+```
+When it is done. A `dump` folder is created in the current directory. The structure of the dump folder is listed below.
+
+```text
+dump
+├── dev
+│   ├── norm
+│   └── raw
+├── test
+│   ├── norm
+│   └── raw
+└── train
+    ├── norm
+    ├── raw
+    └── feats_stats.npy
+```
+
+The dataset is split into 3 parts, namely `train`, `dev` and `test`, each of which contains a `norm` and `raw` subfolder. The `raw` folder contains log magnitude of mel spectrogram of each utterances, while the norm folder contains normalized spectrogram. The statistics used to normalize the spectrogram is computed from the training set, which is located in `dump/train/feats_stats.npy`.
+
+Also there is a `metadata.jsonl` in each subfolder. It is a table-like file which contains id and paths to spectrogam of each utterance.
+
+### Train the model
+```bash
+CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${conf_path} ${train_output_path}
+```
+`./local/train.sh` calls `${BIN_DIR}/train.py`.
+Here's the complete help message.
+
+```text
+usage: train.py [-h] [--config CONFIG] [--train-metadata TRAIN_METADATA]
+                [--dev-metadata DEV_METADATA] [--output-dir OUTPUT_DIR]
+                [--ngpu NGPU] [--verbose VERBOSE] [--batch-size BATCH_SIZE]
+                [--max-iter MAX_ITER] [--run-benchmark RUN_BENCHMARK]
+                [--profiler_options PROFILER_OPTIONS]
+
+Train a ParallelWaveGAN model.
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --config CONFIG       config file to overwrite default config.
+  --train-metadata TRAIN_METADATA
+                        training data.
+  --dev-metadata DEV_METADATA
+                        dev data.
+  --output-dir OUTPUT_DIR
+                        output dir.
+  --ngpu NGPU           if ngpu == 0, use cpu.
+  --verbose VERBOSE     verbose.
+
+benchmark:
+  arguments related to benchmark.
+
+  --batch-size BATCH_SIZE
+                        batch size.
+  --max-iter MAX_ITER   train max steps.
+  --run-benchmark RUN_BENCHMARK
+                        runing benchmark or not, if True, use the --batch-size
+                        and --max-iter.
+  --profiler_options PROFILER_OPTIONS
+                        The option of profiler, which should be in format
+                        "key1=value1;key2=value2;key3=value3".
+```
+
+1. `--config` is a config file in yaml format to overwrite the default config, which can be found at `conf/default.yaml`.
+2. `--train-metadata` and `--dev-metadata` should be the metadata file in the normalized subfolder of `train` and `dev` in the `dump` folder.
+3. `--output-dir` is the directory to save the results of the experiment. Checkpoints are save in `checkpoints/` inside this directory.
+4. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
+
+### Synthesize
+`./local/synthesize.sh` calls `${BIN_DIR}/synthesize.py`, which can synthesize waveform from `metadata.jsonl`.
+```bash
+CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name}
+```
+```text
+usage: synthesize.py [-h] [--config CONFIG] [--checkpoint CHECKPOINT]
+                     [--test-metadata TEST_METADATA] [--output-dir OUTPUT_DIR]
+                     [--ngpu NGPU] [--verbose VERBOSE]
+
+Synthesize with parallel wavegan.
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --config CONFIG       parallel wavegan config file.
+  --checkpoint CHECKPOINT
+                        snapshot to load.
+  --test-metadata TEST_METADATA
+                        dev data.
+  --output-dir OUTPUT_DIR
+                        output dir.
+  --ngpu NGPU           if ngpu == 0, use cpu.
+  --verbose VERBOSE     verbose.
+```
+
+1. `--config` parallel wavegan config file. You should use the same config with which the model is trained.
+2. `--checkpoint` is the checkpoint to load. Pick one of the checkpoints from `checkpoints` inside the training output directory. If you use the pretrained model, use the `snapshot_iter_1000000.pdz `.
+3. `--test-metadata` is the metadata of the test dataset. Use the `metadata.jsonl` in the `dev/norm` subfolder from the processed directory.
+4. `--output-dir` is the directory to save the synthesized audio files.
+5. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
+
+## Pretrained Models
+Pretrained models can be downloaded here [pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_aishell3_ckpt_0.5.zip).
+
+Parallel WaveGAN checkpoint contains files listed below.
+
+```text
+pwg_aishell3_ckpt_0.5
+├── default.yaml                   # default config used to train parallel wavegan
+├── feats_stats.npy                # statistics used to normalize spectrogram when training parallel wavegan
+└── snapshot_iter_1000000.pdz      # generator parameters of parallel wavegan
+```
+## Acknowledgement
+We adapted some code from https://github.com/kan-bayashi/ParallelWaveGAN.

examples/csmsc/tts2/README.md

@@ -7,7 +7,7 @@ Download CSMSC from it's [Official Website](https://test.data-baker.com/data/ind
 
 ### Get MFA result of CSMSC and Extract it
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for SPEEDYSPEECH.
-You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
+You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/BZNSYP`.
@@ -89,7 +89,7 @@ optional arguments:
 6. `--tones-dict` is the path of the tone vocabulary file.
 
 ### Synthesize
-We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
 Download pretrained parallel wavegan model from [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip) and unzip it.
 ```bash
 unzip pwg_baker_ckpt_0.4.zip
@@ -209,6 +209,7 @@ CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path}
 
 ## Pretrained Model
 Pretrained SpeedySpeech model with no silence in the edge of audios. [speedyspeech_nosil_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_ckpt_0.5.zip)
+Static model can be downloaded here [speedyspeech_nosil_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/speedyspeech_nosil_baker_static_0.5.zip).
 
 SpeedySpeech checkpoint contains files listed below.
 ```text

examples/csmsc/tts3/README.md

@@ -7,7 +7,7 @@ Download CSMSC from it's [Official Website](https://test.data-baker.com/data/ind
 
 ### Get MFA result of CSMSC and Extract it
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
-You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
+You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/BZNSYP`.
@@ -87,7 +87,7 @@ optional arguments:
 5. `--phones-dict` is the path of the phone vocabulary file.
 
 ### Synthesize
-We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/csmsc/voc1) as the neural vocoder.
 Download pretrained parallel wavegan model from [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip) and unzip it.
 ```bash
 unzip pwg_baker_ckpt_0.4.zip
@@ -200,6 +200,7 @@ CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path}
 
 ## Pretrained Model
 Pretrained FastSpeech2 model with no silence in the edge of audios. [fastspeech2_nosil_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_ckpt_0.4.zip)
+Static model can be downloaded here [fastspeech2_nosil_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/fastspeech2_nosil_baker_static_0.4.zip)
 
 FastSpeech2 checkpoint contains files listed below.
 ```text

examples/csmsc/voc1/README.md

@@ -6,7 +6,7 @@ Download CSMSC from the [official website](https://www.data-baker.com/data/index
 
 ### Get MFA results for silence trim
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to  cut silence in the edge of audio.
-You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
+You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), or train your own MFA model reference to  [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/BZNSYP`.
@@ -122,7 +122,8 @@ optional arguments:
 5. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
 
 ## Pretrained Models
-Pretrained models can be downloaded here [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip).
+Pretrained model can be downloaded here [pwg_baker_ckpt_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_ckpt_0.4.zip).
+Static models can be downloaded here [pwg_baker_static_0.4.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_baker_static_0.4.zip).
 
 Parallel WaveGAN checkpoint contains files listed below.
 

examples/csmsc/voc3/README.md

@@ -85,11 +85,11 @@ usage: synthesize.py [-h] [--config CONFIG] [--checkpoint CHECKPOINT]
                      [--test-metadata TEST_METADATA] [--output-dir OUTPUT_DIR]
                      [--ngpu NGPU] [--verbose VERBOSE]
 
-Synthesize with parallel wavegan.
+Synthesize with multi band melgan.
 
 optional arguments:
   -h, --help            show this help message and exit
-  --config CONFIG       parallel wavegan config file.
+  --config CONFIG       multi band melgan config file.
   --checkpoint CHECKPOINT
                         snapshot to load.
   --test-metadata TEST_METADATA
@@ -100,10 +100,23 @@ optional arguments:
   --verbose VERBOSE     verbose.
 ```
 
-1. `--config` parallel wavegan config file. You should use the same config with which the model is trained.
+1. `--config` multi band melgan config file. You should use the same config with which the model is trained.
 2. `--checkpoint` is the checkpoint to load. Pick one of the checkpoints from `checkpoints` inside the training output directory.
 3. `--test-metadata` is the metadata of the test dataset. Use the `metadata.jsonl` in the `dev/norm` subfolder from the processed directory.
 4. `--output-dir` is the directory to save the synthesized audio files.
 5. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
 
 ## Pretrained Models
+Pretrained model can be downloaded here [mb_melgan_baker_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_ckpt_0.5.zip).
+Static model can be downloaded here [mb_melgan_baker_static_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/mb_melgan_baker_static_0.5.zip)
+
+Multi Band MelGAN checkpoint contains files listed below.
+
+```text
+mb_melgan_baker_ckpt_0.5
+├── default.yaml                  # default config used to train multi band melgan
+├── feats_stats.npy               # statistics used to normalize spectrogram when training multi band melgan
+└── snapshot_iter_1000000.pdz     # generator parameters of multi band melgan
+```
+## Acknowledgement
+We adapted some code from https://github.com/kan-bayashi/ParallelWaveGAN.

examples/ljspeech/tts0/README.md

@@ -75,7 +75,7 @@ optional arguments:
                         config, passing in KEY VALUE pairs
   -v, --verbose         print msg
 ```
-**Ps.** You can  use [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder to synthesize mels to wavs. (Please  refer to `synthesize.sh` in our  LJSpeech waveflow example)
+**Ps.** You can  use [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder to synthesize mels to wavs. (Please  refer to `synthesize.sh` in our  LJSpeech waveflow example)
 
 ## Pretrained Models
 Pretrained Models can be downloaded from links below. We provide 2 models with different configurations.

examples/ljspeech/tts1/README.md

@@ -78,7 +78,7 @@ optional arguments:
 5. `--phones-dict` is the path of the phone vocabulary file.
 
 ## Synthesize
-We use [waveflow](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder.
+We use [waveflow](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc0) as the neural vocoder.
 Download Pretrained WaveFlow Model with residual channel equals 128 from [waveflow_ljspeech_ckpt_0.3.zip](https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_ljspeech_ckpt_0.3.zip) and unzip it.
 ```bash
 unzip waveflow_ljspeech_ckpt_0.3.zip

examples/ljspeech/tts3/README.md

@@ -7,7 +7,7 @@ Download LJSpeech-1.1 from the [official website](https://keithito.com/LJ-Speech
 
 ### Get MFA result of LJSpeech-1.1 and Extract it
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
-You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
+You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.
@@ -86,7 +86,7 @@ optional arguments:
 5. `--phones-dict` is the path of the phone vocabulary file.
 
 ### Synthesize
-We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/ljspeech/voc1) as the neural vocoder.
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/ljspeech/voc1) as the neural vocoder.
 Download pretrained parallel wavegan model from [pwg_ljspeech_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_ljspeech_ckpt_0.5.zip) and unzip it.
 ```bash
 unzip pwg_ljspeech_ckpt_0.5.zip

examples/ljspeech/voc1/README.md

@@ -5,7 +5,7 @@ This example contains code used to train a [parallel wavegan](http://arxiv.org/a
 Download LJSpeech-1.1 from the [official website](https://keithito.com/LJ-Speech-Dataset/).
 ### Get MFA results for silence trim
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to  cut silence in the edge of audio.
-You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
+You can download from here [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
 
 ## Get Started
 Assume the path to the dataset is `~/datasets/LJSpeech-1.1`.

examples/other/ge2e/README.md

 # Speaker Encoder
-This experiment trains a speaker encoder with speaker verification as its task. It is done as a part of the experiment of transfer learning from speaker verification to multispeaker text-to-speech synthesis, which can be found at [examples/aishell3/vc0](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/aishell3/vc0). The trained speaker encoder is used to extract utterance embeddings from utterances.
+This experiment trains a speaker encoder with speaker verification as its task. It is done as a part of the experiment of transfer learning from speaker verification to multispeaker text-to-speech synthesis, which can be found at [examples/aishell3/vc0](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/vc0). The trained speaker encoder is used to extract utterance embeddings from utterances.
 ## Model
 The model used in this experiment is the speaker encoder with text independent speaker verification task in [GENERALIZED END-TO-END LOSS FOR SPEAKER VERIFICATION](https://arxiv.org/pdf/1710.10467.pdf). GE2E-softmax loss is used.
 

examples/vctk/tts3/README.md

@@ -7,8 +7,8 @@ Download VCTK-0.92 from the [official website](https://datashare.ed.ac.uk/handle
 
 ### Get MFA result of VCTK and Extract it
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for fastspeech2.
-You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
-ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
+You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
+ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
 1. `p315`, because no txt for it.
 2. `p280` and `p362`, because no *_mic2.flac (which is better than *_mic1.flac) for  them.
 
@@ -88,7 +88,7 @@ optional arguments:
 4. `--phones-dict` is the path of the phone vocabulary file.
 
 ### Synthesize
-We use [parallel wavegan](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/vctk/voc1) as the neural vocoder.
+We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/vctk/voc1) as the neural vocoder.
 
 Download pretrained parallel wavegan model from [pwg_vctk_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/pwg_vctk_ckpt_0.5.zip)and unzip it.
 ```bash

examples/vctk/voc1/README.md

@@ -7,8 +7,8 @@ Download VCTK-0.92  from the [official website](https://datashare.ed.ac.uk/handl
 
 ### Get MFA results for silence trim
 We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) results to  cut silence in the edge of audio.
-You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/DeepSpeech/tree/develop/examples/other/use_mfa) of our repo.
-ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
+You can download from here [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz), or train your own MFA model reference to [use_mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/use_mfa) of our repo.
+ps: we remove three speakers in VCTK-0.92 (see [reorganize_vctk.py](https://github.com/PaddlePaddle/PaddleSpeech/blob/develop/examples/other/use_mfa/local/reorganize_vctk.py)):
 1. `p315`, because no txt for it.
 2. `p280` and `p362`, because no *_mic2.flac (which is better than *_mic1.flac) for  them.
 

paddlespeech/t2s/exps/fastspeech2/multi_spk_synthesize_e2e.py

@@ -87,26 +87,27 @@ def evaluate(args, fastspeech2_config, pwg_config):
     output_dir = Path(args.output_dir)
     output_dir.mkdir(parents=True, exist_ok=True)
     # only test the number 0 speaker
-    spk_id = 0
-    for utt_id, sentence in sentences:
-        input_ids = frontend.get_input_ids(sentence, merge_sentences=True)
-        phone_ids = input_ids["phone_ids"]
-        flags = 0
-        for part_phone_ids in phone_ids:
-            with paddle.no_grad():
-                mel = fastspeech2_inference(
-                    part_phone_ids, spk_id=paddle.to_tensor(spk_id))
-                temp_wav = pwg_inference(mel)
-            if flags == 0:
-                wav = temp_wav
-                flags = 1
-            else:
-                wav = paddle.concat([wav, temp_wav])
-        sf.write(
-            str(output_dir / (str(spk_id) + "_" + utt_id + ".wav")),
-            wav.numpy(),
-            samplerate=fastspeech2_config.fs)
-        print(f"{spk_id}_{utt_id} done!")
+    spk_ids = list(range(20))
+    for spk_id in spk_ids:
+        for utt_id, sentence in sentences[:2]:
+            input_ids = frontend.get_input_ids(sentence, merge_sentences=True)
+            phone_ids = input_ids["phone_ids"]
+            flags = 0
+            for part_phone_ids in phone_ids:
+                with paddle.no_grad():
+                    mel = fastspeech2_inference(
+                        part_phone_ids, spk_id=paddle.to_tensor(spk_id))
+                    temp_wav = pwg_inference(mel)
+                if flags == 0:
+                    wav = temp_wav
+                    flags = 1
+                else:
+                    wav = paddle.concat([wav, temp_wav])
+            sf.write(
+                str(output_dir / (str(spk_id) + "_" + utt_id + ".wav")),
+                wav.numpy(),
+                samplerate=fastspeech2_config.fs)
+            print(f"{spk_id}_{utt_id} done!")
 
 
 def main():

paddlespeech/t2s/exps/gan_vocoder/multi_band_melgan/synthesize.py

@@ -30,9 +30,9 @@ from paddlespeech.t2s.models.melgan import MelGANGenerator
 
 def main():
     parser = argparse.ArgumentParser(
-        description="Synthesize with parallel wavegan.")
+        description="Synthesize with multi band melgan.")
     parser.add_argument(
-        "--config", type=str, help="parallel wavegan config file.")
+        "--config", type=str, help="multi band melgan config file.")
     parser.add_argument("--checkpoint", type=str, help="snapshot to load.")
     parser.add_argument("--test-metadata", type=str, help="dev data.")
     parser.add_argument("--output-dir", type=str, help="output dir.")

paddlespeech/t2s/frontend/zh_frontend.py

@@ -219,18 +219,45 @@ class Frontend():
     def get_phonemes(self,
                      sentence: str,
                      merge_sentences: bool=True,
-                     with_erhua: bool=True) -> List[List[str]]:
+                     with_erhua: bool=True,
+                     robot: bool=False,
+                     print_info: bool=False) -> List[List[str]]:
         sentences = self.text_normalizer.normalize(sentence)
         phonemes = self._g2p(
             sentences, merge_sentences=merge_sentences, with_erhua=with_erhua)
+        # change all tones to `1`
+        if robot:
+            new_phonemes = []
+            for sentence in phonemes:
+                new_sentence = []
+                for item in sentence:
+                    # `er` only have tone `2`
+                    if item[-1] in "12345" and item != "er2":
+                        item = item[:-1] + "1"
+                    new_sentence.append(item)
+                new_phonemes.append(new_sentence)
+            phonemes = new_phonemes
+        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,
-            get_tone_ids: bool=False) -> Dict[str, List[paddle.Tensor]]:
-        phonemes = self.get_phonemes(sentence, merge_sentences=merge_sentences)
+    def get_input_ids(self,
+                      sentence: str,
+                      merge_sentences: bool=True,
+                      get_tone_ids: bool=False,
+                      robot: bool=False,
+                      print_info: bool=False) -> Dict[str, List[paddle.Tensor]]:
+        phonemes = self.get_phonemes(
+            sentence,
+            merge_sentences=merge_sentences,
+            print_info=print_info,
+            robot=robot)
         result = {}
         phones = []
         tones = []

paddlespeech/t2s/models/fastspeech2/fastspeech2.py

@@ -513,9 +513,9 @@ class FastSpeech2(nn.Layer):
         spembs : Tensor, optional
             peaker embedding vector (spk_embed_dim,).
         spk_id : Tensor, optional(int64)
-            Speaker embedding vector (spk_embed_dim).
+            Batch of padded spk ids  (1,).
         tone_id : Tensor, optional(int64)
-            Batch of padded tone ids  (B, Tmax).
+            Batch of padded tone ids  (T,).
 
         Returns
         ----------
@@ -526,9 +526,7 @@ class FastSpeech2(nn.Layer):
         x = paddle.cast(text, 'int64')
         y = speech
         spemb = spembs
-        if durations is not None:
-            d = paddle.cast(durations, 'int64')
-        p, e = pitch, energy
+        d, p, e = durations, pitch, energy
         # setup batch axis
         ilens = paddle.shape(x)[0]
 
@@ -539,8 +537,9 @@ class FastSpeech2(nn.Layer):
 
         if spemb is not None:
             spembs = spemb.unsqueeze(0)
-        else:
-            spembs = None
+
+        if tone_id is not None:
+            tone_id = tone_id.unsqueeze(0)
 
         if use_teacher_forcing:
             # use groundtruth of duration, pitch, and energy
@@ -549,7 +548,7 @@ class FastSpeech2(nn.Layer):
             es = e.unsqueeze(0) if e is not None else None
             # ds, ps, es = , p.unsqueeze(0), e.unsqueeze(0)
             # (1, L, odim)
-            _, outs, d_outs, *_ = self._forward(
+            _, outs, d_outs, p_outs, e_outs = self._forward(
                 xs,
                 ilens,
                 ys,
@@ -562,7 +561,7 @@ class FastSpeech2(nn.Layer):
                 is_inference=True)
         else:
             # (1, L, odim)
-            _, outs, d_outs, *_ = self._forward(
+            _, outs, d_outs, p_outs, e_outs = self._forward(
                 xs,
                 ilens,
                 ys,
@@ -571,8 +570,7 @@ class FastSpeech2(nn.Layer):
                 spembs=spembs,
                 spk_id=spk_id,
                 tone_id=tone_id)
-
-        return outs[0]
+        return outs[0], d_outs[0], p_outs[0], e_outs[0]
 
     def _integrate_with_spk_embed(self, hs, spembs):
         """Integrate speaker embedding with hidden states.
@@ -683,7 +681,8 @@ class FastSpeech2Inference(nn.Layer):
         self.acoustic_model = model
 
     def forward(self, text, spk_id=None):
-        normalized_mel = self.acoustic_model.inference(text, spk_id=spk_id)
+        normalized_mel, d_outs, p_outs, e_outs = self.acoustic_model.inference(
+            text, spk_id=spk_id)
         logmel = self.normalizer.inverse(normalized_mel)
         return logmel
 

paddlespeech/t2s/modules/fastspeech2_transformer/encoder.py

@@ -12,8 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # Modified from espnet(https://github.com/espnet/espnet)
-import logging
-
 from paddle import nn
 
 from paddlespeech.t2s.modules.fastspeech2_transformer.attention import MultiHeadedAttention
@@ -122,7 +120,6 @@ class Encoder(nn.Layer):
                 "rel_selfattn",
                 "legacy_rel_selfattn",
         ]:
-            logging.info("encoder self-attention layer type = self-attention")
             encoder_selfattn_layer = MultiHeadedAttention
             encoder_selfattn_layer_args = [
                 (attention_heads, attention_dim, attention_dropout_rate, )

tools/Makefile

@@ -51,7 +51,7 @@ soxbindings.done:
 	touch soxbindings.done
 
 mfa.done:
-	test -d montreal-forced-aligner || $(WGET) https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner/releases/download/v1.0.1/montreal-forced-aligner_linux.tar.gz
+	test -d montreal-forced-aligner || $(WGET) https://paddlespeech.bj.bcebos.com/Parakeet/montreal-forced-aligner_linux.tar.gz
 	tar xvf montreal-forced-aligner_linux.tar.gz
 	touch mfa.done