msb-public
/
PaddleSpeech
mirror of https://github.com/PaddlePaddle/PaddleSpeech
1
0
Code Issues Projects Releases Wiki Activity

add zh_en mix example, test=tts

Browse Source
pull/2234/head
lym0302 2 years ago
parent 070a08f2be
commit 894556f871
17 changed files with 810 additions and 75 deletions
Show Stats Download Patch File Download Diff File

280
examples/zh_en_tts/tts3/README.md
Unescape View File

@ -1,26 +1,272 @@
# Test # Mixed Chinese and English TTS with CSMSC, LJSpeech-1.1, AISHELL-3 and VCTK datasets
We train a Chinese-English mixed fastspeech2 model. The training code is still being sorted out, let's show how to use it first.
The sample rate of the synthesized audio is 22050 Hz.
## Download pretrained models This example contains code used to train a [Fastspeech2](https://arxiv.org/abs/2006.04558) model with [CSMSC](https://www.data-baker.com/open_source.html), [LJSpeech-1.1](https://keithito.com/LJ-Speech-Dataset/), [AISHELL3](http://www.aishelltech.com/aishell_3) and [VCTK](https://datashare.ed.ac.uk/handle/10283/3443) datasets.
Put pretrained models in a directory named `models`.
- [fastspeech2_csmscljspeech_add-zhen.zip](https://paddlespeech.bj.bcebos.com/t2s/chinse_english_mixed/models/fastspeech2_csmscljspeech_add-zhen.zip)
- [hifigan_ljspeech_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_ckpt_0.2.0.zip) ## Dataset
### Download and Extract
Download all datasets and extract it to `~/datasets`. The CSMSC dataset is in the directory `~/datasets/BZNSYP`. The Ljspeech dataset is in the directory `~/datasets/LJSpeech-1.1`. The aishell3 dataset is in the directory `~/datasets/data_aishell3`. The vctk dataset is in the directory `~/datasets/VCTK-Corpus-0.92`.
### Get MFA Result and Extract
We use [MFA](https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner) to get durations for the fastspeech2 training.
You can download from here [baker_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/BZNSYP/with_tone/baker_alignment_tone.tar.gz), [ljspeech_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/LJSpeech-1.1/ljspeech_alignment.tar.gz), [aishell3_alignment_tone.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/AISHELL-3/with_tone/aishell3_alignment_tone.tar.gz) and [vctk_alignment.tar.gz](https://paddlespeech.bj.bcebos.com/MFA/VCTK-Corpus-0.92/vctk_alignment.tar.gz). Or train your MFA model reference to [mfa example](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/other/mfa) (use MFA1.x now) of our repo.
## Get Started
Assume the paths of the datasets are `~/datasets/BZNSYP`, `~/datasets/LJSpeech-1.1`, `~/datasets/data_aishell3` and `~/datasets/VCTK-Corpus-0.92`.
Assume the path to the MFA result of the datasets are `./mfa_results/baker_alignment_tone`, `./mfa_results/ljspeech_alignment`, `./mfa_results/aishell3_alignment_tone` and `./mfa_results/vctk_alignment`.
Run the command below to
1. **source path**.
2. preprocess the dataset.
3. train the model.
4. synthesize wavs.
- synthesize waveform from `metadata.jsonl`.
- synthesize waveform from text file.
```bash
./run.sh
```
You can choose a range of stages you want to run, or set `stage` equal to `stop-stage` to use only one stage, for example, running the following command will only preprocess the dataset.
```bash
./run.sh --stage 0 --stop-stage 0
```
### Data Preprocessing
```bash
./local/preprocess.sh ${conf_path} ${datasets_root_dir} ${mfa_root_dir}
```
When it is done. A `dump` folder is created in the current directory. The structure of the dump folder is listed below.
```text
dump
├── dev
│ ├── norm
│ └── raw
├── phone_id_map.txt
├── speaker_id_map.txt
├── test
│ ├── norm
│ └── raw
└── train
├── energy_stats.npy
├── norm
├── pitch_stats.npy
├── raw
└── speech_stats.npy
```
The dataset is split into 3 parts, namely `train`, `dev`, and` test`, each of which contains a `norm` and `raw` subfolder. The raw folder contains speech, pitch and energy features of each utterance, while the norm folder contains normalized ones. The statistics used to normalize features are computed from the training set, which is located in `dump/train/*_stats.npy`.
Also, there is a `metadata.jsonl` in each subfolder. It is a table-like file that contains phones, text_lengths, speech_lengths, durations, the path of speech features, the path of pitch features, a path of energy features, speaker, and id of each utterance.
### Model Training
`./local/train.sh` calls `${BIN_DIR}/train.py`.
```bash
CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${conf_path} ${train_output_path}
```
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] [--phones-dict PHONES_DICT]
[--speaker-dict SPEAKER_DICT] [--voice-cloning VOICE_CLONING]
Train a FastSpeech2 model.
optional arguments:
-h, --help show this help message and exit
--config CONFIG fastspeech2 config file.
--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.
--phones-dict PHONES_DICT
phone vocabulary file.
--speaker-dict SPEAKER_DICT
speaker id map file for multiple speaker model.
--voice-cloning VOICE_CLONING
whether training voice cloning model.
```
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 saved in `checkpoints/` inside this directory.
4. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
5. `--phones-dict` is the path of the phone vocabulary file.
6. `--speaker-dict` is the path of the speaker id map file when training a multi-speaker FastSpeech2.
### Synthesizing
We use [parallel wavegan](https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/examples/aishell3/voc1) as the neural vocoder.
Download the pretrained parallel wavegan model from [pwg_aishell3_ckpt_0.5.zip](https://paddlespeech.bj.bcebos.com/Parakeet/released_models/pwgan/pwg_aishell3_ckpt_0.5.zip) and unzip it.
```bash ```bash
mkdir models unzip pwg_aishell3_ckpt_0.5.zip
cd models
wget https://paddlespeech.bj.bcebos.com/t2s/chinse_english_mixed/models/fastspeech2_csmscljspeech_add-zhen.zip
unzip fastspeech2_csmscljspeech_add-zhen.zip
wget https://paddlespeech.bj.bcebos.com/Parakeet/released_models/hifigan/hifigan_ljspeech_ckpt_0.2.0.zip
unzip hifigan_ljspeech_ckpt_0.2.0.zip
cd ../
``` ```
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
```
`./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]
[--am {speedyspeech_csmsc,fastspeech2_csmsc,fastspeech2_ljspeech,fastspeech2_aishell3,fastspeech2_vctk,tacotron2_csmsc,tacotron2_ljspeech,tacotron2_aishell3, fastspeech2_mix}]
[--am_config AM_CONFIG] [--am_ckpt AM_CKPT]
[--am_stat AM_STAT] [--phones_dict PHONES_DICT]
[--tones_dict TONES_DICT] [--speaker_dict SPEAKER_DICT]
[--voice-cloning VOICE_CLONING]
[--voc {pwgan_csmsc,pwgan_ljspeech,pwgan_aishell3,pwgan_vctk,mb_melgan_csmsc,wavernn_csmsc,hifigan_csmsc,hifigan_ljspeech,hifigan_aishell3,hifigan_vctk,style_melgan_csmsc}]
[--voc_config VOC_CONFIG] [--voc_ckpt VOC_CKPT]
[--voc_stat VOC_STAT] [--ngpu NGPU]
[--test_metadata TEST_METADATA] [--output_dir OUTPUT_DIR]
## test Synthesize with acoustic model & vocoder
You can choose `--spk_id` {0, 1} in `local/synthesize_e2e.sh`.
optional arguments:
-h, --help show this help message and exit
--am {speedyspeech_csmsc,fastspeech2_csmsc,fastspeech2_ljspeech,fastspeech2_aishell3,fastspeech2_vctk,tacotron2_csmsc,tacotron2_ljspeech,tacotron2_aishell3, fastspeech2_mix}
Choose acoustic model type of tts task.
--am_config AM_CONFIG
Config of acoustic model.
--am_ckpt AM_CKPT Checkpoint file of acoustic model.
--am_stat AM_STAT mean and standard deviation used to normalize
spectrogram when training acoustic model.
--phones_dict PHONES_DICT
phone vocabulary file.
--tones_dict TONES_DICT
tone vocabulary file.
--speaker_dict SPEAKER_DICT
speaker id map file.
--voice-cloning VOICE_CLONING
whether training voice cloning model.
--voc {pwgan_csmsc,pwgan_ljspeech,pwgan_aishell3,pwgan_vctk,mb_melgan_csmsc,wavernn_csmsc,hifigan_csmsc,hifigan_ljspeech,hifigan_aishell3,hifigan_vctk,style_melgan_csmsc}
Choose vocoder type of tts task.
--voc_config VOC_CONFIG
Config of voc.
--voc_ckpt VOC_CKPT Checkpoint file of voc.
--voc_stat VOC_STAT mean and standard deviation used to normalize
spectrogram when training voc.
--ngpu NGPU if ngpu == 0, use cpu.
--test_metadata TEST_METADATA
test metadata.
--output_dir OUTPUT_DIR
output dir.
```
`./local/synthesize_e2e.sh` calls `${BIN_DIR}/../synthesize_e2e.py`, which can synthesize waveform from text file.
```bash ```bash
bash test.sh CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name}
```
```text
usage: synthesize_e2e.py [-h]
[--am {speedyspeech_csmsc,speedyspeech_aishell3,fastspeech2_csmsc,fastspeech2_ljspeech,fastspeech2_aishell3,fastspeech2_vctk,tacotron2_csmsc,tacotron2_ljspeech, fastspeech2_mix}]
[--am_config AM_CONFIG] [--am_ckpt AM_CKPT]
[--am_stat AM_STAT] [--phones_dict PHONES_DICT]
[--tones_dict TONES_DICT]
[--speaker_dict SPEAKER_DICT] [--spk_id SPK_ID]
[--voc {pwgan_csmsc,pwgan_ljspeech,pwgan_aishell3,pwgan_vctk,mb_melgan_csmsc,style_melgan_csmsc,hifigan_csmsc,hifigan_ljspeech,hifigan_aishell3,hifigan_vctk,wavernn_csmsc}]
[--voc_config VOC_CONFIG] [--voc_ckpt VOC_CKPT]
[--voc_stat VOC_STAT] [--lang LANG]
[--inference_dir INFERENCE_DIR] [--ngpu NGPU]
[--text TEXT] [--output_dir OUTPUT_DIR]
Synthesize with acoustic model & vocoder
optional arguments:
-h, --help show this help message and exit
--am {speedyspeech_csmsc,speedyspeech_aishell3,fastspeech2_csmsc,fastspeech2_ljspeech,fastspeech2_aishell3,fastspeech2_vctk,tacotron2_csmsc,tacotron2_ljspeech, fastspeech2_mix}
Choose acoustic model type of tts task.
--am_config AM_CONFIG
Config of acoustic model.
--am_ckpt AM_CKPT Checkpoint file of acoustic model.
--am_stat AM_STAT mean and standard deviation used to normalize
spectrogram when training acoustic model.
--phones_dict PHONES_DICT
phone vocabulary file.
--tones_dict TONES_DICT
tone vocabulary file.
--speaker_dict SPEAKER_DICT
speaker id map file.
--spk_id SPK_ID spk id for multi speaker acoustic model
--voc {pwgan_csmsc,pwgan_ljspeech,pwgan_aishell3,pwgan_vctk,mb_melgan_csmsc,style_melgan_csmsc,hifigan_csmsc,hifigan_ljspeech,hifigan_aishell3,hifigan_vctk,wavernn_csmsc}
Choose vocoder type of tts task.
--voc_config VOC_CONFIG
Config of voc.
--voc_ckpt VOC_CKPT Checkpoint file of voc.
--voc_stat VOC_STAT mean and standard deviation used to normalize
spectrogram when training voc.
--lang LANG Choose model language. zh or en or mix
--inference_dir INFERENCE_DIR
dir to save inference models
--ngpu NGPU if ngpu == 0, use cpu.
--text TEXT text to synthesize, a 'utt_id sentence' pair per line.
--output_dir OUTPUT_DIR
output dir.
```
1. `--am` is acoustic model type with the format {model_name}_{dataset}
2. `--am_config`, `--am_ckpt`, `--am_stat`, `--phones_dict` `--speaker_dict` are arguments for acoustic model, which correspond to the 5 files in the fastspeech2 pretrained model.
3. `--voc` is vocoder type with the format {model_name}_{dataset}
4. `--voc_config`, `--voc_ckpt`, `--voc_stat` are arguments for vocoder, which correspond to the 3 files in the parallel wavegan pretrained model.
5. `--lang` is the model language, which can be `zh` or `en` or `mix`.
6. `--test_metadata` should be the metadata file in the normalized subfolder of `test` in the `dump` folder.
7. `--text` is the text file, which contains sentences to synthesize.
8. `--output_dir` is the directory to save synthesized audio files.
9. `--ngpu` is the number of gpus to use, if ngpu == 0, use cpu.
## Pretrained Model
Pretrained FastSpeech2 model with no silence in the edge of audios:
- [fastspeech2_mix_ckpt_0.2.0.zip](https://paddlespeech.bj.bcebos.com/t2s/chinse_english_mixed/models/fastspeech2_mix_ckpt_0.2.0.zip)
The static model can be downloaded here:
- [fastspeech2_mix_static_0.2.0.zip](https://paddlespeech.bj.bcebos.com/t2s/chinse_english_mixed/models/fastspeech2_mix_static_0.2.0.zip)
The ONNX model can be downloaded here:
- [fastspeech2_mix_onnx_0.2.0.zip](https://paddlespeech.bj.bcebos.com/t2s/chinse_english_mixed/models/fastspeech2_mix_onnx_0.2.0.zip)
FastSpeech2 checkpoint contains files listed below.
```text
fastspeech2_mix_ckpt_0.2.0
├── default.yaml # default config used to train fastspeech2
├── phone_id_map.txt # phone vocabulary file when training fastspeech2
├── snapshot_iter_99200.pdz # model parameters and optimizer states
├── speaker_id_map.txt # speaker id map file when training a multi-speaker fastspeech2
└── speech_stats.npy # statistics used to normalize spectrogram when training fastspeech2
```
You can use the following scripts to synthesize for `${BIN_DIR}/../sentences_mix.txt` using pretrained fastspeech2 and parallel wavegan models.
174 means baker speaker, 175 means ljspeech speaker. For other speaker information, see `speaker_id_map.txt`
```bash
source path.sh
FLAGS_allocator_strategy=naive_best_fit \
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
python3 ${BIN_DIR}/../synthesize_e2e.py \
--am=fastspeech2_mix \
--am_config=fastspeech2_mix_ckpt_0.2.0/default.yaml \
--am_ckpt=fastspeech2_mix_ckpt_0.2.0/snapshot_iter_96400.pdz \
--am_stat=fastspeech2_mix_ckpt_0.2.0/speech_stats.npy \
--voc=pwgan_aishell3 \
--voc_config=pwg_aishell3_ckpt_0.5/default.yaml \
--voc_ckpt=pwg_aishell3_ckpt_0.5/snapshot_iter_1000000.pdz \
--voc_stat=pwg_aishell3_ckpt_0.5/feats_stats.npy \
--lang=mix \
--text=${BIN_DIR}/../sentences_mix.txt \
--output_dir=exp/default/test_e2e \
--phones_dict=fastspeech2_mix_ckpt_0.2.0/phone_id_map.txt \
--speaker_dict=fastspeech2_mix_ckpt_0.2.0/speaker_id_map.txt \
--spk_id=174 \
--inference_dir=exp/default/inference
``` ```

104
examples/zh_en_tts/tts3/conf/default.yaml
Unescape View File

@ -0,0 +1,104 @@
###########################################################
# FEATURE EXTRACTION SETTING #
###########################################################
fs: 24000 # sr
n_fft: 2048 # FFT size (samples).
n_shift: 300 # Hop size (samples). 12.5ms
win_length: 1200 # Window length (samples). 50ms
# If set to null, it will be the same as fft_size.
window: "hann" # Window function.
# Only used for feats_type != raw
fmin: 80 # Minimum frequency of Mel basis.
fmax: 7600 # Maximum frequency of Mel basis.
n_mels: 80 # The number of mel basis.
# Only used for the model using pitch features (e.g. FastSpeech2)
f0min: 80 # Minimum f0 for pitch extraction.
f0max: 400 # Maximum f0 for pitch extraction.
###########################################################
# DATA SETTING #
###########################################################
batch_size: 64
num_workers: 2
###########################################################
# MODEL SETTING #
###########################################################
model:
adim: 384 # attention dimension
aheads: 2 # number of attention heads
elayers: 4 # number of encoder layers
eunits: 1536 # number of encoder ff units
dlayers: 4 # number of decoder layers
dunits: 1536 # number of decoder ff units
positionwise_layer_type: conv1d # type of position-wise layer
positionwise_conv_kernel_size: 3 # kernel size of position wise conv layer
duration_predictor_layers: 2 # number of layers of duration predictor
duration_predictor_chans: 256 # number of channels of duration predictor
duration_predictor_kernel_size: 3 # filter size of duration predictor
postnet_layers: 5 # number of layers of postnset
postnet_filts: 5 # filter size of conv layers in postnet
postnet_chans: 256 # number of channels of conv layers in postnet
use_scaled_pos_enc: True # whether to use scaled positional encoding
encoder_normalize_before: True # whether to perform layer normalization before the input
decoder_normalize_before: True # whether to perform layer normalization before the input
reduction_factor: 1 # reduction factor
init_type: xavier_uniform # initialization type
init_enc_alpha: 1.0 # initial value of alpha of encoder scaled position encoding
init_dec_alpha: 1.0 # initial value of alpha of decoder scaled position encoding
transformer_enc_dropout_rate: 0.2 # dropout rate for transformer encoder layer
transformer_enc_positional_dropout_rate: 0.2 # dropout rate for transformer encoder positional encoding
transformer_enc_attn_dropout_rate: 0.2 # dropout rate for transformer encoder attention layer
transformer_dec_dropout_rate: 0.2 # dropout rate for transformer decoder layer
transformer_dec_positional_dropout_rate: 0.2 # dropout rate for transformer decoder positional encoding
transformer_dec_attn_dropout_rate: 0.2 # dropout rate for transformer decoder attention layer
pitch_predictor_layers: 5 # number of conv layers in pitch predictor
pitch_predictor_chans: 256 # number of channels of conv layers in pitch predictor
pitch_predictor_kernel_size: 5 # kernel size of conv leyers in pitch predictor
pitch_predictor_dropout: 0.5 # dropout rate in pitch predictor
pitch_embed_kernel_size: 1 # kernel size of conv embedding layer for pitch
pitch_embed_dropout: 0.0 # dropout rate after conv embedding layer for pitch
stop_gradient_from_pitch_predictor: True # whether to stop the gradient from pitch predictor to encoder
energy_predictor_layers: 2 # number of conv layers in energy predictor
energy_predictor_chans: 256 # number of channels of conv layers in energy predictor
energy_predictor_kernel_size: 3 # kernel size of conv leyers in energy predictor
energy_predictor_dropout: 0.5 # dropout rate in energy predictor
energy_embed_kernel_size: 1 # kernel size of conv embedding layer for energy
energy_embed_dropout: 0.0 # dropout rate after conv embedding layer for energy
stop_gradient_from_energy_predictor: False # whether to stop the gradient from energy predictor to encoder
spk_embed_dim: 256 # speaker embedding dimension
spk_embed_integration_type: concat # speaker embedding integration type
###########################################################
# UPDATER SETTING #
###########################################################
updater:
use_masking: True # whether to apply masking for padded part in loss calculation
###########################################################
# OPTIMIZER SETTING #
###########################################################
optimizer:
optim: adam # optimizer type
learning_rate: 0.001 # learning rate
###########################################################
# TRAINING SETTING #
###########################################################
max_epoch: 200
num_snapshots: 5
###########################################################
# OTHER SETTING #
###########################################################
seed: 10086

39
examples/zh_en_tts/tts3/local/inference.sh
Unescape View File

@ -0,0 +1,39 @@
#!/bin/bash
train_output_path=$1
stage=0
stop_stage=0
# voc: pwgan_aishell3
# the spk_id=174 means baker speaker, default
# the spk_id=175 means ljspeech speaker
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_mix \
--voc=pwgan_aishell3 \
--text=${BIN_DIR}/../sentences_mix.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt \
--lang=mix \
--spk_id=174
fi
# voc: hifigan_aishell3
# the spk_id=174 means baker speaker, default
# the spk_id=175 means ljspeech speaker
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../inference.py \
--inference_dir=${train_output_path}/inference \
--am=fastspeech2_mix \
--voc=hifigan_aishell3 \
--text=${BIN_DIR}/../sentences_mix.txt \
--output_dir=${train_output_path}/pd_infer_out \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt \
--lang=mix \
--spk_id=174
fi

43
examples/zh_en_tts/tts3/local/ort_predict.sh
Unescape View File

@ -0,0 +1,43 @@
train_output_path=$1
stage=0
stop_stage=0
# e2e, synthesize from text
# voc: pwgan_aishell3
# the spk_id=174 means baker speaker, default
# the spk_id=175 means ljspeech speaker
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_mix \
--voc=pwgan_aishell3 \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_mix.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=4 \
--lang=mix \
--spk_id=174
fi
# voc: hifigan_aishell3
# the spk_id=174 means baker speaker, default
# the spk_id=175 means ljspeech speaker
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
python3 ${BIN_DIR}/../ort_predict_e2e.py \
--inference_dir=${train_output_path}/inference_onnx \
--am=fastspeech2_mix \
--voc=hifigan_aishell3 \
--output_dir=${train_output_path}/onnx_infer_out_e2e \
--text=${BIN_DIR}/../sentences_mix.txt \
--phones_dict=dump/phone_id_map.txt \
--device=cpu \
--cpu_threads=4 \
--lang=mix \
--spk_id=174
fi

1
examples/zh_en_tts/tts3/local/paddle2onnx.sh
Unescape View File

@ -0,0 +1 @@
../../../csmsc/tts3/local/paddle2onnx.sh

149
examples/zh_en_tts/tts3/local/preprocess.sh
Unescape View File

@ -0,0 +1,149 @@
#!/bin/bash
stage=0
stop_stage=100
config_path=$1
datasets_root_dir=$2
mfa_root_dir=$3
# 1. get durations from MFA's result
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
echo "Generate durations_baker.txt from MFA results ..."
python3 ${MAIN_ROOT}/utils/gen_duration_from_textgrid.py \
--inputdir=${mfa_root_dir}/baker_alignment_tone \
--output durations_baker.txt \
--config=${config_path}
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
echo "Generate durations_ljspeech.txt from MFA results ..."
python3 ${MAIN_ROOT}/utils/gen_duration_from_textgrid.py \
--inputdir=${mfa_root_dir}/ljspeech_alignment \
--output durations_ljspeech.txt \
--config=${config_path}
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
echo "Generate durations_aishell3.txt from MFA results ..."
python3 ${MAIN_ROOT}/utils/gen_duration_from_textgrid.py \
--inputdir=${mfa_root_dir}/aishell3_alignment_tone \
--output durations_aishell3.txt \
--config=${config_path}
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
echo "Generate durations_vctk.txt from MFA results ..."
python3 ${MAIN_ROOT}/utils/gen_duration_from_textgrid.py \
--inputdir=${mfa_root_dir}/vctk_alignment \
--output durations_vctk.txt \
--config=${config_path}
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# concat duration file
echo "concat durations_baker.txt, durations_ljspeech.txt, durations_aishell3.txt and durations_vctk.txt to durations.txt"
cat durations_baker.txt durations_ljspeech.txt durations_aishell3.txt durations_vctk.txt > durations.txt
fi
# 2. extract features
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
echo "Extract baker features ..."
python3 ${BIN_DIR}/preprocess.py \
--dataset=baker \
--rootdir=${datasets_root_dir}/BZNSYP/ \
--dumpdir=dump \
--dur-file=durations.txt \
--config=${config_path} \
--num-cpu=20 \
--cut-sil=True \
--write_metadata_method=a
fi
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
echo "Extract ljspeech features ..."
python3 ${BIN_DIR}/preprocess.py \
--dataset=ljspeech \
--rootdir=${datasets_root_dir}/LJSpeech-1.1/ \
--dumpdir=dump \
--dur-file=durations.txt \
--config=${config_path} \
--num-cpu=20 \
--cut-sil=True \
--write_metadata_method=a
fi
if [ ${stage} -le 7 ] && [ ${stop_stage} -ge 7 ]; then
echo "Extract aishell3 features ..."
python3 ${BIN_DIR}/preprocess.py \
--dataset=aishell3 \
--rootdir=${datasets_root_dir}/data_aishell3/ \
--dumpdir=dump \
--dur-file=durations.txt \
--config=${config_path} \
--num-cpu=20 \
--cut-sil=True \
--write_metadata_method=a
fi
if [ ${stage} -le 8 ] && [ ${stop_stage} -ge 8 ]; then
echo "Extract vctk features ..."
python3 ${BIN_DIR}/preprocess.py \
--dataset=vctk \
--rootdir=${datasets_root_dir}/VCTK-Corpus-0.92/ \
--dumpdir=dump \
--dur-file=durations.txt \
--config=${config_path} \
--num-cpu=20 \
--cut-sil=True \
--write_metadata_method=a
fi
# 3. get features' stats(mean and std)
if [ ${stage} -le 9 ] && [ ${stop_stage} -ge 9 ]; then
echo "Get features' stats ..."
python3 ${MAIN_ROOT}/utils/compute_statistics.py \
--metadata=dump/train/raw/metadata.jsonl \
--field-name="speech"
python3 ${MAIN_ROOT}/utils/compute_statistics.py \
--metadata=dump/train/raw/metadata.jsonl \
--field-name="pitch"
python3 ${MAIN_ROOT}/utils/compute_statistics.py \
--metadata=dump/train/raw/metadata.jsonl \
--field-name="energy"
fi
# 4. normalize and covert phone/speaker to id, dev and test should use train's stats
if [ ${stage} -le 10 ] && [ ${stop_stage} -ge 10 ]; then
echo "Normalize ..."
python3 ${BIN_DIR}/normalize.py \
--metadata=dump/train/raw/metadata.jsonl \
--dumpdir=dump/train/norm \
--speech-stats=dump/train/speech_stats.npy \
--pitch-stats=dump/train/pitch_stats.npy \
--energy-stats=dump/train/energy_stats.npy \
--phones-dict=dump/phone_id_map.txt \
--speaker-dict=dump/speaker_id_map.txt
python3 ${BIN_DIR}/normalize.py \
--metadata=dump/dev/raw/metadata.jsonl \
--dumpdir=dump/dev/norm \
--speech-stats=dump/train/speech_stats.npy \
--pitch-stats=dump/train/pitch_stats.npy \
--energy-stats=dump/train/energy_stats.npy \
--phones-dict=dump/phone_id_map.txt \
--speaker-dict=dump/speaker_id_map.txt
python3 ${BIN_DIR}/normalize.py \
--metadata=dump/test/raw/metadata.jsonl \
--dumpdir=dump/test/norm \
--speech-stats=dump/train/speech_stats.npy \
--pitch-stats=dump/train/pitch_stats.npy \
--energy-stats=dump/train/energy_stats.npy \
--phones-dict=dump/phone_id_map.txt \
--speaker-dict=dump/speaker_id_map.txt
fi

50
examples/zh_en_tts/tts3/local/synthesize.sh
Unescape View File

@ -0,0 +1,50 @@
#!/bin/bash
config_path=$1
train_output_path=$2
ckpt_name=$3
stage=0
stop_stage=0
# voc: pwgan_aishell3
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 ${BIN_DIR}/../synthesize.py \
--am=fastspeech2_mix \
--am_config=${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 \
--test_metadata=dump/test/norm/metadata2.jsonl \
--output_dir=${train_output_path}/test \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt
fi
# voc: hifigan_aishell3
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
FLAGS_allocator_strategy=naive_best_fit \
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
python3 ${BIN_DIR}/../synthesize.py \
--am=fastspeech2_mix \
--am_config=${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 \
--test_metadata=dump/test/norm/metadata.jsonl \
--output_dir=${train_output_path}/test \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt
fi

64
examples/zh_en_tts/tts3/local/synthesize_e2e.sh
Unescape View File

@ -1,31 +1,57 @@
#!/bin/bash #!/bin/bash
model_dir=$1 config_path=$1
output=$2 train_output_path=$2
am_name=fastspeech2_csmscljspeech_add-zhen ckpt_name=$3
am_model_dir=${model_dir}/${am_name}/
stage=1 stage=0
stop_stage=1 stop_stage=0
# voc: pwgan_aishell3
# the spk_id=174 means baker speaker, default.
# the spk_id=175 means ljspeech speaker
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 ${BIN_DIR}/../synthesize_e2e.py \
--am=fastspeech2_mix \
--am_config=${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=mix \
--text=${BIN_DIR}/../sentences_mix.txt \
--output_dir=${train_output_path}/test_e2e \
--phones_dict=dump/phone_id_map.txt \
--speaker_dict=dump/speaker_id_map.txt \
--spk_id=174 \
--inference_dir=${train_output_path}/inference
fi
# hifigan # voc: hifigan_aishell3
# the spk_id=174 means baker speaker, default
# the spk_id=175 means ljspeech speaker
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
echo "in hifigan syn_e2e"
FLAGS_allocator_strategy=naive_best_fit \ FLAGS_allocator_strategy=naive_best_fit \
FLAGS_fraction_of_gpu_memory_to_use=0.01 \ FLAGS_fraction_of_gpu_memory_to_use=0.01 \
python3 ${BIN_DIR}/../synthesize_e2e.py \ python3 ${BIN_DIR}/../synthesize_e2e.py \
--am=fastspeech2_mix \ --am=fastspeech2_mix \
--am_config=${am_model_dir}/default.yaml \ --am_config=${config_path} \
--am_ckpt=${am_model_dir}/snapshot_iter_94000.pdz \ --am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
--am_stat=${am_model_dir}/speech_stats.npy \ --am_stat=dump/train/speech_stats.npy \
--voc=hifigan_ljspeech \ --voc=hifigan_aishell3 \
--voc_config=${model_dir}/hifigan_ljspeech_ckpt_0.2.0/default.yaml \ --voc_config=hifigan_aishell3_ckpt_0.2.0/default.yaml \
--voc_ckpt=${model_dir}/hifigan_ljspeech_ckpt_0.2.0/snapshot_iter_2500000.pdz \ --voc_ckpt=hifigan_aishell3_ckpt_0.2.0/snapshot_iter_2500000.pdz \
--voc_stat=${model_dir}/hifigan_ljspeech_ckpt_0.2.0/feats_stats.npy \ --voc_stat=hifigan_aishell3_ckpt_0.2.0/feats_stats.npy \
--lang=mix \ --lang=mix \
--text=${BIN_DIR}/../sentences_mix.txt \ --text=${BIN_DIR}/../sentences_mix.txt \
--output_dir=${output}/test_e2e \ --output_dir=${train_output_path}/test_e2e \
--phones_dict=${am_model_dir}/phone_id_map.txt \ --phones_dict=dump/phone_id_map.txt \
--speaker_dict=${am_model_dir}/speaker_id_map.txt \ --speaker_dict=dump/speaker_id_map.txt \
--spk_id 0 --spk_id=174 \
fi --inference_dir=${train_output_path}/inference
fi

13
examples/zh_en_tts/tts3/local/train.sh
Unescape View File

@ -0,0 +1,13 @@
#!/bin/bash
config_path=$1
train_output_path=$2
python3 ${BIN_DIR}/train.py \
--train-metadata=dump/train/norm/metadata.jsonl \
--dev-metadata=dump/dev/norm/metadata.jsonl \
--config=${config_path} \
--output-dir=${train_output_path} \
--ngpu=2 \
--phones-dict=dump/phone_id_map.txt \
--speaker-dict=dump/speaker_id_map.txt

63
examples/zh_en_tts/tts3/run.sh
Unescape View File

@ -0,0 +1,63 @@
#!/bin/bash
set -e
source path.sh
gpus=0,1
stage=0
stop_stage=0
datasets_root_dir=./datasets/
mfa_root_dir=./mfa_results/
conf_path=conf/default.yaml
train_output_path=exp/default
ckpt_name=snapshot_iter_99200.pdz
# with the following command, you can choose 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
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
# prepare data
./local/preprocess.sh ${conf_path} ${datasets_root_dir} ${mfa_root_dir} || exit -1
fi
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
# train model, all `ckpt` under `train_output_path/checkpoints/` dir
CUDA_VISIBLE_DEVICES=${gpus} ./local/train.sh ${conf_path} ${train_output_path} || exit -1
fi
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
# synthesize, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${conf_path} ${train_output_path} ${ckpt_name} || exit -1
fi
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
# inference with static model, vocoder is pwgan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/inference.sh ${train_output_path} || exit -1
fi
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
# install paddle2onnx
version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
if [[ -z "$version" || ${version} != '0.9.8' ]]; then
pip install paddle2onnx==0.9.8
fi
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_mix
# considering the balance between speed and quality, we recommend that you use hifigan as vocoder
./local/paddle2onnx.sh ${train_output_path} inference inference_onnx pwgan_aishell3
#./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_aishell3
fi
# inference with onnxruntime, use fastspeech2 + hifigan by default
if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
./local/ort_predict.sh ${train_output_path}
fi

23
examples/zh_en_tts/tts3/test.sh
Unescape View File

@ -1,23 +0,0 @@
#!/bin/bash
set -e
source path.sh
gpus=0,1
stage=3
stop_stage=100
model_dir=models
output_dir=output
# with the following command, you can choose 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
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
# synthesize_e2e, vocoder is hifigan by default
CUDA_VISIBLE_DEVICES=${gpus} ./local/synthesize_e2e.sh ${model_dir} ${output_dir} || exit -1
fi

24
paddlespeech/t2s/exps/fastspeech2/preprocess.py
Unescape View File

@ -144,7 +144,8 @@ def process_sentences(config,
energy_extractor=None, energy_extractor=None,
nprocs: int=1, nprocs: int=1,
cut_sil: bool=True, cut_sil: bool=True,
spk_emb_dir: Path=None): spk_emb_dir: Path=None,
write_metadata_method: str='w'):
if nprocs == 1: if nprocs == 1:
results = [] results = []
for fp in tqdm.tqdm(fps, total=len(fps)): for fp in tqdm.tqdm(fps, total=len(fps)):
@ -179,7 +180,10 @@ def process_sentences(config,
results.append(record) results.append(record)
results.sort(key=itemgetter("utt_id")) results.sort(key=itemgetter("utt_id"))
with jsonlines.open(output_dir / "metadata.jsonl", 'w') as writer: print("wwwwwwwwwwwwwwwwwwwwwwwwwwwrite_metadata_method: ",
write_metadata_method)
with jsonlines.open(output_dir / "metadata.jsonl",
write_metadata_method) as writer:
for item in results: for item in results:
writer.write(item) writer.write(item)
print("Done") print("Done")
@ -223,6 +227,13 @@ def main():
default=None, default=None,
type=str, type=str,
help="directory to speaker embedding files.") help="directory to speaker embedding files.")
parser.add_argument(
"--write_metadata_method",
default="w",
type=str,
choices=["w", "a"],
help="How the metadata.jsonl file is written.")
args = parser.parse_args() args = parser.parse_args()
rootdir = Path(args.rootdir).expanduser() rootdir = Path(args.rootdir).expanduser()
@ -340,7 +351,8 @@ def main():
energy_extractor=energy_extractor, energy_extractor=energy_extractor,
nprocs=args.num_cpu, nprocs=args.num_cpu,
cut_sil=args.cut_sil, cut_sil=args.cut_sil,
spk_emb_dir=spk_emb_dir) spk_emb_dir=spk_emb_dir,
write_metadata_method=args.write_metadata_method)
if dev_wav_files: if dev_wav_files:
process_sentences( process_sentences(
config=config, config=config,
@ -351,7 +363,8 @@ def main():
pitch_extractor=pitch_extractor, pitch_extractor=pitch_extractor,
energy_extractor=energy_extractor, energy_extractor=energy_extractor,
cut_sil=args.cut_sil, cut_sil=args.cut_sil,
spk_emb_dir=spk_emb_dir) spk_emb_dir=spk_emb_dir,
write_metadata_method=args.write_metadata_method)
if test_wav_files: if test_wav_files:
process_sentences( process_sentences(
config=config, config=config,
@ -363,7 +376,8 @@ def main():
energy_extractor=energy_extractor, energy_extractor=energy_extractor,
nprocs=args.num_cpu, nprocs=args.num_cpu,
cut_sil=args.cut_sil, cut_sil=args.cut_sil,
spk_emb_dir=spk_emb_dir) spk_emb_dir=spk_emb_dir,
write_metadata_method=args.write_metadata_method)
if __name__ == "__main__": if __name__ == "__main__":

9
paddlespeech/t2s/exps/inference.py
Unescape View File

@ -41,6 +41,7 @@ def parse_args():
'fastspeech2_ljspeech', 'fastspeech2_ljspeech',
'fastspeech2_vctk', 'fastspeech2_vctk',
'tacotron2_csmsc', 'tacotron2_csmsc',
'fastspeech2_mix',
], ],
help='Choose acoustic model type of tts task.') help='Choose acoustic model type of tts task.')
parser.add_argument( parser.add_argument(
@ -77,7 +78,7 @@ def parse_args():
'--lang', '--lang',
type=str, type=str,
default='zh', default='zh',
help='Choose model language. zh or en') help='Choose model language. zh or en or mix')
parser.add_argument( parser.add_argument(
"--text", "--text",
type=str, type=str,
@ -156,7 +157,8 @@ def main():
frontend=frontend, frontend=frontend,
lang=args.lang, lang=args.lang,
merge_sentences=merge_sentences, merge_sentences=merge_sentences,
speaker_dict=args.speaker_dict, ) speaker_dict=args.speaker_dict,
spk_id=args.spk_id, )
wav = get_voc_output( wav = get_voc_output(
voc_predictor=voc_predictor, input=am_output_data) voc_predictor=voc_predictor, input=am_output_data)
speed = wav.size / t.elapse speed = wav.size / t.elapse
@ -178,7 +180,8 @@ def main():
frontend=frontend, frontend=frontend,
lang=args.lang, lang=args.lang,
merge_sentences=merge_sentences, merge_sentences=merge_sentences,
speaker_dict=args.speaker_dict, ) speaker_dict=args.speaker_dict,
spk_id=args.spk_id, )
wav = get_voc_output( wav = get_voc_output(
voc_predictor=voc_predictor, input=am_output_data) voc_predictor=voc_predictor, input=am_output_data)

11
paddlespeech/t2s/exps/ort_predict_e2e.py
Unescape View File

@ -76,7 +76,7 @@ def ort_predict(args):
else: else:
phone_ids = np.random.randint(1, 266, size=(T, )) phone_ids = np.random.randint(1, 266, size=(T, ))
am_input_feed.update({'text': phone_ids}) am_input_feed.update({'text': phone_ids})
if am_dataset in {"aishell3", "vctk"}: if am_dataset in {"aishell3", "vctk", "mix"}:
am_input_feed.update({'spk_id': spk_id}) am_input_feed.update({'spk_id': spk_id})
elif am_name == 'speedyspeech': elif am_name == 'speedyspeech':
@ -112,13 +112,17 @@ def ort_predict(args):
input_ids = frontend.get_input_ids( input_ids = frontend.get_input_ids(
sentence, merge_sentences=merge_sentences) sentence, merge_sentences=merge_sentences)
phone_ids = input_ids["phone_ids"] phone_ids = input_ids["phone_ids"]
elif args.lang == 'mix':
input_ids = frontend.get_input_ids(
sentence, merge_sentences=merge_sentences)
phone_ids = input_ids["phone_ids"]
else: else:
print("lang should in {'zh', 'en'}!") print("lang should in {'zh', 'en', 'mix'}!")
# merge_sentences=True here, so we only use the first item of phone_ids # merge_sentences=True here, so we only use the first item of phone_ids
phone_ids = phone_ids[0].numpy() phone_ids = phone_ids[0].numpy()
if am_name == 'fastspeech2': if am_name == 'fastspeech2':
am_input_feed.update({'text': phone_ids}) am_input_feed.update({'text': phone_ids})
if am_dataset in {"aishell3", "vctk"}: if am_dataset in {"aishell3", "vctk", "mix"}:
am_input_feed.update({'spk_id': spk_id}) am_input_feed.update({'spk_id': spk_id})
elif am_name == 'speedyspeech': elif am_name == 'speedyspeech':
tone_ids = tone_ids[0].numpy() tone_ids = tone_ids[0].numpy()
@ -154,6 +158,7 @@ def parse_args():
'fastspeech2_ljspeech', 'fastspeech2_ljspeech',
'fastspeech2_vctk', 'fastspeech2_vctk',
'speedyspeech_csmsc', 'speedyspeech_csmsc',
'fastspeech2_mix',
], ],
help='Choose acoustic model type of tts task.') help='Choose acoustic model type of tts task.')
parser.add_argument( parser.add_argument(

8
paddlespeech/t2s/exps/syn_utils.py
Unescape View File

@ -273,7 +273,8 @@ def am_to_static(am_inference,
am_name = am[:am.rindex('_')] am_name = am[:am.rindex('_')]
am_dataset = am[am.rindex('_') + 1:] am_dataset = am[am.rindex('_') + 1:]
if am_name == 'fastspeech2': if am_name == 'fastspeech2':
if am_dataset in {"aishell3", "vctk"} and speaker_dict is not None: if am_dataset in {"aishell3", "vctk", "mix"
} and speaker_dict is not None:
am_inference = jit.to_static( am_inference = jit.to_static(
am_inference, am_inference,
input_spec=[ input_spec=[
@ -285,7 +286,8 @@ def am_to_static(am_inference,
am_inference, input_spec=[InputSpec([-1], dtype=paddle.int64)]) am_inference, input_spec=[InputSpec([-1], dtype=paddle.int64)])
elif am_name == 'speedyspeech': elif am_name == 'speedyspeech':
if am_dataset in {"aishell3", "vctk"} and speaker_dict is not None: if am_dataset in {"aishell3", "vctk", "mix"
} and speaker_dict is not None:
am_inference = jit.to_static( am_inference = jit.to_static(
am_inference, am_inference,
input_spec=[ input_spec=[
@ -356,7 +358,7 @@ def get_am_output(
get_spk_id = False get_spk_id = False
if am_name == 'speedyspeech': if am_name == 'speedyspeech':
get_tone_ids = True get_tone_ids = True
if am_dataset in {"aishell3", "vctk"} and speaker_dict: if am_dataset in {"aishell3", "vctk", "mix"} and speaker_dict:
get_spk_id = True get_spk_id = True
spk_id = np.array([spk_id]) spk_id = np.array([spk_id])
if lang == 'zh': if lang == 'zh':

2
paddlespeech/t2s/exps/synthesize.py
Unescape View File

@ -136,7 +136,7 @@ def parse_args():
choices=[ choices=[
'speedyspeech_csmsc', 'fastspeech2_csmsc', 'fastspeech2_ljspeech', 'speedyspeech_csmsc', 'fastspeech2_csmsc', 'fastspeech2_ljspeech',
'fastspeech2_aishell3', 'fastspeech2_vctk', 'tacotron2_csmsc', 'fastspeech2_aishell3', 'fastspeech2_vctk', 'tacotron2_csmsc',
'tacotron2_ljspeech', 'tacotron2_aishell3' 'tacotron2_ljspeech', 'tacotron2_aishell3', 'fastspeech2_mix'
], ],
help='Choose acoustic model type of tts task.') help='Choose acoustic model type of tts task.')
parser.add_argument( parser.add_argument(

2
paddlespeech/t2s/exps/synthesize_e2e.py
Unescape View File

@ -133,7 +133,7 @@ def evaluate(args):
mel = am_inference(part_phone_ids) mel = am_inference(part_phone_ids)
elif am_name == 'speedyspeech': elif am_name == 'speedyspeech':
part_tone_ids = tone_ids[i] part_tone_ids = tone_ids[i]
if am_dataset in {"aishell3", "vctk"}: if am_dataset in {"aishell3", "vctk", "mix"}:
spk_id = paddle.to_tensor(args.spk_id) spk_id = paddle.to_tensor(args.spk_id)
mel = am_inference(part_phone_ids, part_tone_ids, mel = am_inference(part_phone_ids, part_tone_ids,
spk_id) spk_id)

Write Preview
Loading…
Cancel
Save