Honei
3 years ago
committed by
GitHub
commit
1395b5f5fa
@ -0,0 +1 @@
|
||||
*.wav
|
||||
@ -1,4 +1,10 @@
|
||||
#!/bin/bash
|
||||
|
||||
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav
|
||||
|
||||
# asr
|
||||
paddlespeech asr --input ./zh.wav
|
||||
|
||||
|
||||
# asr + punc
|
||||
paddlespeech asr --input ./zh.wav | paddlespeech text --task punc
|
||||
@ -1,7 +1,8 @@
|
||||
model: 'conformer_wenetspeech'
|
||||
lang: 'zh'
|
||||
sample_rate: 16000
|
||||
cfg_path:
|
||||
ckpt_path:
|
||||
cfg_path: # [optional]
|
||||
ckpt_path: # [optional]
|
||||
decode_method: 'attention_rescoring'
|
||||
force_yes: False
|
||||
force_yes: True
|
||||
device: # set 'gpu:id' or 'cpu'
|
||||
|
||||
@ -0,0 +1,26 @@
|
||||
# This is the parameter configuration file for ASR server.
|
||||
# These are the static models that support paddle inference.
|
||||
|
||||
##################################################################
|
||||
# ACOUSTIC MODEL SETTING #
|
||||
# am choices=['deepspeech2offline_aishell'] TODO
|
||||
##################################################################
|
||||
model_type: 'deepspeech2offline_aishell'
|
||||
am_model: # the pdmodel file of am static model [optional]
|
||||
am_params: # the pdiparams file of am static model [optional]
|
||||
lang: 'zh'
|
||||
sample_rate: 16000
|
||||
cfg_path:
|
||||
decode_method:
|
||||
force_yes: True
|
||||
|
||||
am_predictor_conf:
|
||||
device: # set 'gpu:id' or 'cpu'
|
||||
switch_ir_optim: True
|
||||
glog_info: False # True -> print glog
|
||||
summary: True # False -> do not show predictor config
|
||||
|
||||
|
||||
##################################################################
|
||||
# OTHERS #
|
||||
##################################################################
|
||||
@ -1,3 +1,7 @@
|
||||
#!/bin/bash
|
||||
|
||||
# single process
|
||||
paddlespeech tts --input 今天的天气不错啊
|
||||
|
||||
# Batch process
|
||||
echo -e "1 欢迎光临。\n2 谢谢惠顾。" | paddlespeech tts
|
||||
@ -0,0 +1,369 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "a1e738e0",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## 获取测试的 logit 数据"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"id": "29d3368b",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"hlens.npy\n",
|
||||
"logits.npy\n",
|
||||
"ys_lens.npy\n",
|
||||
"ys_pad.npy\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"!mkdir -p ./test_data\n",
|
||||
"!test -f ./test_data/ctc_loss_compare_data.tgz || wget -P ./test_data https://paddlespeech.bj.bcebos.com/datasets/unit_test/asr/ctc_loss_compare_data.tgz\n",
|
||||
"!tar xzvf test_data/ctc_loss_compare_data.tgz -C ./test_data\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"id": "240caf1d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os\n",
|
||||
"import numpy as np\n",
|
||||
"import time\n",
|
||||
"\n",
|
||||
"data_dir=\"./test_data\"\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"id": "91bad949",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"logits_np = np.load(os.path.join(data_dir, \"logits.npy\"))\n",
|
||||
"ys_pad_np = np.load(os.path.join(data_dir, \"ys_pad.npy\"))\n",
|
||||
"hlens_np = np.load(os.path.join(data_dir, \"hlens.npy\"))\n",
|
||||
"ys_lens_np = np.load(os.path.join(data_dir, \"ys_lens.npy\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4cef2f15",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## 使用 torch 的 ctc loss"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 4,
|
||||
"id": "90612004",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"'1.10.1+cu102'"
|
||||
]
|
||||
},
|
||||
"execution_count": 4,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"import torch\n",
|
||||
"torch.__version__"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 5,
|
||||
"id": "00799f97",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"def torch_ctc_loss(use_cpu):\n",
|
||||
" if use_cpu:\n",
|
||||
" device = torch.device(\"cpu\")\n",
|
||||
" else:\n",
|
||||
" device = torch.device(\"cuda\")\n",
|
||||
"\n",
|
||||
" reduction_type = \"sum\" \n",
|
||||
"\n",
|
||||
" ctc_loss = torch.nn.CTCLoss(reduction=reduction_type)\n",
|
||||
"\n",
|
||||
" ys_hat = torch.tensor(logits_np, device = device)\n",
|
||||
" ys_pad = torch.tensor(ys_pad_np, device = device)\n",
|
||||
" hlens = torch.tensor(hlens_np, device = device)\n",
|
||||
" ys_lens = torch.tensor(ys_lens_np, device = device)\n",
|
||||
"\n",
|
||||
" ys_hat = ys_hat.transpose(0, 1)\n",
|
||||
" \n",
|
||||
" # 开始计算时间\n",
|
||||
" start_time = time.time()\n",
|
||||
" ys_hat = ys_hat.log_softmax(2)\n",
|
||||
" loss = ctc_loss(ys_hat, ys_pad, hlens, ys_lens)\n",
|
||||
" end_time = time.time()\n",
|
||||
" \n",
|
||||
" loss = loss / ys_hat.size(1)\n",
|
||||
" return end_time - start_time, loss.item()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "ba47b5a4",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## 使用 paddle 的 ctc loss"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"id": "6882a06e",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"data": {
|
||||
"text/plain": [
|
||||
"'2.2.2'"
|
||||
]
|
||||
},
|
||||
"execution_count": 6,
|
||||
"metadata": {},
|
||||
"output_type": "execute_result"
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"import paddle\n",
|
||||
"paddle.__version__"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 7,
|
||||
"id": "3cfa3b7c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"def paddle_ctc_loss(use_cpu): \n",
|
||||
" import paddle.nn as pn\n",
|
||||
" if use_cpu:\n",
|
||||
" device = \"cpu\"\n",
|
||||
" else:\n",
|
||||
" device = \"gpu\"\n",
|
||||
"\n",
|
||||
" paddle.set_device(device)\n",
|
||||
"\n",
|
||||
" logits = paddle.to_tensor(logits_np)\n",
|
||||
" ys_pad = paddle.to_tensor(ys_pad_np,dtype='int32')\n",
|
||||
" hlens = paddle.to_tensor(hlens_np, dtype='int64')\n",
|
||||
" ys_lens = paddle.to_tensor(ys_lens_np, dtype='int64')\n",
|
||||
"\n",
|
||||
" logits = logits.transpose([1,0,2])\n",
|
||||
"\n",
|
||||
" ctc_loss = pn.CTCLoss(reduction='sum')\n",
|
||||
" # 开始计算时间\n",
|
||||
" start_time = time.time()\n",
|
||||
" pn_loss = ctc_loss(logits, ys_pad, hlens, ys_lens)\n",
|
||||
" end_time = time.time()\n",
|
||||
" \n",
|
||||
" pn_loss = pn_loss / logits.shape[1]\n",
|
||||
" return end_time - start_time, pn_loss.item()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 8,
|
||||
"id": "40413ef9",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"CPU, iteration 10\n",
|
||||
"torch_ctc_loss 159.17137145996094\n",
|
||||
"paddle_ctc_loss 159.16574096679688\n",
|
||||
"paddle average time 1.718252992630005\n",
|
||||
"torch average time 0.17536230087280275\n",
|
||||
"paddle time / torch time (cpu) 9.798303193320452\n",
|
||||
"\n",
|
||||
"GPU, iteration 10\n",
|
||||
"torch_ctc_loss 159.172119140625\n",
|
||||
"paddle_ctc_loss 159.17205810546875\n",
|
||||
"paddle average time 0.018606925010681154\n",
|
||||
"torch average time 0.0026710033416748047\n",
|
||||
"paddle time / torch time (gpu) 6.966267963938231\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# 使用 CPU\n",
|
||||
"\n",
|
||||
"iteration = 10\n",
|
||||
"use_cpu = True\n",
|
||||
"torch_total_time = 0\n",
|
||||
"paddle_total_time = 0\n",
|
||||
"for _ in range(iteration):\n",
|
||||
" cost_time, torch_loss = torch_ctc_loss(use_cpu)\n",
|
||||
" torch_total_time += cost_time\n",
|
||||
"for _ in range(iteration):\n",
|
||||
" cost_time, paddle_loss = paddle_ctc_loss(use_cpu)\n",
|
||||
" paddle_total_time += cost_time\n",
|
||||
"print (\"CPU, iteration\", iteration)\n",
|
||||
"print (\"torch_ctc_loss\", torch_loss)\n",
|
||||
"print (\"paddle_ctc_loss\", paddle_loss)\n",
|
||||
"print (\"paddle average time\", paddle_total_time / iteration)\n",
|
||||
"print (\"torch average time\", torch_total_time / iteration)\n",
|
||||
"print (\"paddle time / torch time (cpu)\" , paddle_total_time/ torch_total_time)\n",
|
||||
"\n",
|
||||
"print (\"\")\n",
|
||||
"\n",
|
||||
"# 使用 GPU\n",
|
||||
"\n",
|
||||
"use_cpu = False\n",
|
||||
"torch_total_time = 0\n",
|
||||
"paddle_total_time = 0\n",
|
||||
"for _ in range(iteration):\n",
|
||||
" cost_time, torch_loss = torch_ctc_loss(use_cpu)\n",
|
||||
" torch_total_time += cost_time\n",
|
||||
"for _ in range(iteration):\n",
|
||||
" cost_time, paddle_loss = paddle_ctc_loss(use_cpu)\n",
|
||||
" paddle_total_time += cost_time\n",
|
||||
"print (\"GPU, iteration\", iteration)\n",
|
||||
"print (\"torch_ctc_loss\", torch_loss)\n",
|
||||
"print (\"paddle_ctc_loss\", paddle_loss)\n",
|
||||
"print (\"paddle average time\", paddle_total_time / iteration)\n",
|
||||
"print (\"torch average time\", torch_total_time / iteration)\n",
|
||||
"print (\"paddle time / torch time (gpu)\" , paddle_total_time/ torch_total_time)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "7cdf8697",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## 其他: 使用 PaddleSpeech 中的 ctcloss 查一下loss值"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 9,
|
||||
"id": "73fad81d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"logits_np = np.load(os.path.join(data_dir, \"logits.npy\"))\n",
|
||||
"ys_pad_np = np.load(os.path.join(data_dir, \"ys_pad.npy\"))\n",
|
||||
"hlens_np = np.load(os.path.join(data_dir, \"hlens.npy\"))\n",
|
||||
"ys_lens_np = np.load(os.path.join(data_dir, \"ys_lens.npy\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 10,
|
||||
"id": "2b41e45d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"2022-02-25 11:34:34.143 | INFO | paddlespeech.s2t.modules.loss:__init__:41 - CTCLoss Loss reduction: sum, div-bs: True\n",
|
||||
"2022-02-25 11:34:34.143 | INFO | paddlespeech.s2t.modules.loss:__init__:42 - CTCLoss Grad Norm Type: instance\n",
|
||||
"2022-02-25 11:34:34.144 | INFO | paddlespeech.s2t.modules.loss:__init__:73 - CTCLoss() kwargs:{'norm_by_times': True}, not support: {'norm_by_batchsize': False, 'norm_by_total_logits_len': False}\n",
|
||||
"loss 159.17205810546875\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"name": "stderr",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"/root/miniconda3/lib/python3.7/site-packages/paddle/fluid/dygraph/math_op_patch.py:253: UserWarning: The dtype of left and right variables are not the same, left dtype is paddle.float32, but right dtype is paddle.int32, the right dtype will convert to paddle.float32\n",
|
||||
" format(lhs_dtype, rhs_dtype, lhs_dtype))\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"use_cpu = False\n",
|
||||
"\n",
|
||||
"from paddlespeech.s2t.modules.loss import CTCLoss\n",
|
||||
"\n",
|
||||
"if use_cpu:\n",
|
||||
" device = \"cpu\"\n",
|
||||
"else:\n",
|
||||
" device = \"gpu\"\n",
|
||||
"\n",
|
||||
"paddle.set_device(device)\n",
|
||||
"\n",
|
||||
"blank_id=0\n",
|
||||
"reduction_type='sum'\n",
|
||||
"batch_average= True\n",
|
||||
"grad_norm_type='instance'\n",
|
||||
"\n",
|
||||
"criterion = CTCLoss(\n",
|
||||
" blank=blank_id,\n",
|
||||
" reduction=reduction_type,\n",
|
||||
" batch_average=batch_average,\n",
|
||||
" grad_norm_type=grad_norm_type)\n",
|
||||
"\n",
|
||||
"logits = paddle.to_tensor(logits_np)\n",
|
||||
"ys_pad = paddle.to_tensor(ys_pad_np,dtype='int32')\n",
|
||||
"hlens = paddle.to_tensor(hlens_np, dtype='int64')\n",
|
||||
"ys_lens = paddle.to_tensor(ys_lens_np, dtype='int64')\n",
|
||||
"\n",
|
||||
"pn_ctc_loss = criterion(logits, ys_pad, hlens, ys_lens)\n",
|
||||
"print(\"loss\", pn_ctc_loss.item())\n",
|
||||
" "
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "de525d38",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## 结论\n",
|
||||
"在 CPU 环境下： torch 的 CTC loss 的计算速度是 paddle 的 9.8 倍 \n",
|
||||
"在 GPU 环境下： torch 的 CTC loss 的计算速度是 paddle 的 6.87 倍\n",
|
||||
"\n",
|
||||
"## 其他结论\n",
|
||||
"torch 的 ctc loss 在 CPU 和 GPU 下 都没有完全对齐。其中CPU的前向对齐精度大约为 1e-2。 GPU 的前向对齐精度大约为 1e-4 。"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.7.10"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@ -0,0 +1,110 @@
|
||||
###########################################################
|
||||
# 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 # Maximum f0 for pitch extraction.
|
||||
f0max: 400 # Minimum f0 for pitch extraction.
|
||||
|
||||
|
||||
###########################################################
|
||||
# DATA SETTING #
|
||||
###########################################################
|
||||
batch_size: 32
|
||||
num_workers: 4
|
||||
|
||||
|
||||
###########################################################
|
||||
# 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
|
||||
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
|
||||
encoder_type: conformer # encoder type
|
||||
decoder_type: conformer # decoder type
|
||||
conformer_pos_enc_layer_type: rel_pos # conformer positional encoding type
|
||||
conformer_self_attn_layer_type: rel_selfattn # conformer self-attention type
|
||||
conformer_activation_type: swish # conformer activation type
|
||||
use_macaron_style_in_conformer: true # whether to use macaron style in conformer
|
||||
use_cnn_in_conformer: true # whether to use CNN in conformer
|
||||
conformer_enc_kernel_size: 7 # kernel size in CNN module of conformer-based encoder
|
||||
conformer_dec_kernel_size: 31 # kernel size in CNN module of conformer-based decoder
|
||||
init_type: xavier_uniform # initialization type
|
||||
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: 1000
|
||||
num_snapshots: 5
|
||||
|
||||
|
||||
###########################################################
|
||||
# OTHER SETTING #
|
||||
###########################################################
|
||||
seed: 10086
|
||||
@ -1,20 +1,105 @@
|
||||
#!/bin/bash
|
||||
|
||||
config_path=$1
|
||||
train_output_path=$2
|
||||
ckpt_name=$3
|
||||
stage=0
|
||||
stop_stage=0
|
||||
|
||||
# pwgan
|
||||
if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
|
||||
FLAGS_allocator_strategy=naive_best_fit \
|
||||
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||
python3 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/speech_stats.npy \
|
||||
--voc=pwgan_csmsc \
|
||||
--voc_config=pwg_baker_ckpt_0.4/pwg_default.yaml \
|
||||
--voc_ckpt=pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
|
||||
--voc_stat=pwg_baker_ckpt_0.4/pwg_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--phones_dict=dump/phone_id_map.txt \
|
||||
--tones_dict=dump/tone_id_map.txt
|
||||
fi
|
||||
|
||||
# for more GAN Vocoders
|
||||
# multi band melgan
|
||||
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 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/speech_stats.npy \
|
||||
--voc=mb_melgan_csmsc \
|
||||
--voc_config=mb_melgan_csmsc_ckpt_0.1.1/default.yaml \
|
||||
--voc_ckpt=mb_melgan_csmsc_ckpt_0.1.1/snapshot_iter_1000000.pdz\
|
||||
--voc_stat=mb_melgan_csmsc_ckpt_0.1.1/feats_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--phones_dict=dump/phone_id_map.txt \
|
||||
--tones_dict=dump/tone_id_map.txt
|
||||
fi
|
||||
|
||||
# style melgan
|
||||
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
|
||||
FLAGS_allocator_strategy=naive_best_fit \
|
||||
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||
python3 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/speech_stats.npy \
|
||||
--voc=style_melgan_csmsc \
|
||||
--voc_config=style_melgan_csmsc_ckpt_0.1.1/default.yaml \
|
||||
--voc_ckpt=style_melgan_csmsc_ckpt_0.1.1/snapshot_iter_1500000.pdz \
|
||||
--voc_stat=style_melgan_csmsc_ckpt_0.1.1/feats_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--phones_dict=dump/phone_id_map.txt \
|
||||
--tones_dict=dump/tone_id_map.txt
|
||||
fi
|
||||
|
||||
# hifigan
|
||||
if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
|
||||
echo "in hifigan syn"
|
||||
FLAGS_allocator_strategy=naive_best_fit \
|
||||
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||
python3 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/speech_stats.npy \
|
||||
--voc=hifigan_csmsc \
|
||||
--voc_config=hifigan_csmsc_ckpt_0.1.1/default.yaml \
|
||||
--voc_ckpt=hifigan_csmsc_ckpt_0.1.1/snapshot_iter_2500000.pdz \
|
||||
--voc_stat=hifigan_csmsc_ckpt_0.1.1/feats_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--phones_dict=dump/phone_id_map.txt \
|
||||
--tones_dict=dump/tone_id_map.txt
|
||||
fi
|
||||
|
||||
FLAGS_allocator_strategy=naive_best_fit \
|
||||
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||
python3 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/feats_stats.npy \
|
||||
--voc=pwgan_csmsc \
|
||||
--voc_config=pwg_baker_ckpt_0.4/pwg_default.yaml \
|
||||
--voc_ckpt=pwg_baker_ckpt_0.4/pwg_snapshot_iter_400000.pdz \
|
||||
--voc_stat=pwg_baker_ckpt_0.4/pwg_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--phones_dict=dump/phone_id_map.txt \
|
||||
--tones_dict=dump/tone_id_map.txt
|
||||
# wavernn
|
||||
if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
|
||||
echo "in wavernn syn"
|
||||
FLAGS_allocator_strategy=naive_best_fit \
|
||||
FLAGS_fraction_of_gpu_memory_to_use=0.01 \
|
||||
python3 ${BIN_DIR}/../synthesize.py \
|
||||
--am=speedyspeech_csmsc \
|
||||
--am_config=${config_path} \
|
||||
--am_ckpt=${train_output_path}/checkpoints/${ckpt_name} \
|
||||
--am_stat=dump/train/speech_stats.npy \
|
||||
--voc=wavernn_csmsc \
|
||||
--voc_config=wavernn_csmsc_ckpt_0.2.0/default.yaml \
|
||||
--voc_ckpt=wavernn_csmsc_ckpt_0.2.0/snapshot_iter_400000.pdz \
|
||||
--voc_stat=wavernn_csmsc_ckpt_0.2.0/feats_stats.npy \
|
||||
--test_metadata=dump/test/norm/metadata.jsonl \
|
||||
--output_dir=${train_output_path}/test \
|
||||
--tones_dict=dump/tone_id_map.txt \
|
||||
--phones_dict=dump/phone_id_map.txt
|
||||
fi
|
||||
|
||||
@ -0,0 +1,14 @@
|
||||
# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
from .infer import StatsExecutor
|
||||
@ -0,0 +1,193 @@
|
||||
# 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 typing import List
|
||||
|
||||
from prettytable import PrettyTable
|
||||
|
||||
from ..log import logger
|
||||
from ..utils import cli_register
|
||||
from ..utils import stats_wrapper
|
||||
|
||||
__all__ = ['StatsExecutor']
|
||||
|
||||
model_name_format = {
|
||||
'asr': 'Model-Language-Sample Rate',
|
||||
'cls': 'Model-Sample Rate',
|
||||
'st': 'Model-Source language-Target language',
|
||||
'text': 'Model-Task-Language',
|
||||
'tts': 'Model-Language'
|
||||
}
|
||||
|
||||
|
||||
@cli_register(
|
||||
name='paddlespeech.stats',
|
||||
description='Get speech tasks support models list.')
|
||||
class StatsExecutor():
|
||||
def __init__(self):
|
||||
super(StatsExecutor, self).__init__()
|
||||
|
||||
self.parser = argparse.ArgumentParser(
|
||||
prog='paddlespeech.stats', add_help=True)
|
||||
self.parser.add_argument(
|
||||
'--task',
|
||||
type=str,
|
||||
default='asr',
|
||||
choices=['asr', 'cls', 'st', 'text', 'tts'],
|
||||
help='Choose speech task.',
|
||||
required=True)
|
||||
self.task_choices = ['asr', 'cls', 'st', 'text', 'tts']
|
||||
|
||||
def show_support_models(self, pretrained_models: dict):
|
||||
fields = model_name_format[self.task].split("-")
|
||||
table = PrettyTable(fields)
|
||||
for key in pretrained_models:
|
||||
table.add_row(key.split("-"))
|
||||
print(table)
|
||||
|
||||
def execute(self, argv: List[str]) -> bool:
|
||||
"""
|
||||
Command line entry.
|
||||
"""
|
||||
parser_args = self.parser.parse_args(argv)
|
||||
self.task = parser_args.task
|
||||
if self.task not in self.task_choices:
|
||||
logger.error(
|
||||
"Please input correct speech task, choices = ['asr', 'cls', 'st', 'text', 'tts']"
|
||||
)
|
||||
return False
|
||||
|
||||
elif self.task == 'asr':
|
||||
try:
|
||||
from ..asr.infer import pretrained_models
|
||||
logger.info(
|
||||
"Here is the list of ASR pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
return True
|
||||
except BaseException:
|
||||
logger.error("Failed to get the list of ASR pretrained models.")
|
||||
return False
|
||||
|
||||
elif self.task == 'cls':
|
||||
try:
|
||||
from ..cls.infer import pretrained_models
|
||||
logger.info(
|
||||
"Here is the list of CLS pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
return True
|
||||
except BaseException:
|
||||
logger.error("Failed to get the list of CLS pretrained models.")
|
||||
return False
|
||||
|
||||
elif self.task == 'st':
|
||||
try:
|
||||
from ..st.infer import pretrained_models
|
||||
logger.info(
|
||||
"Here is the list of ST pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
return True
|
||||
except BaseException:
|
||||
logger.error("Failed to get the list of ST pretrained models.")
|
||||
return False
|
||||
|
||||
elif self.task == 'text':
|
||||
try:
|
||||
from ..text.infer import pretrained_models
|
||||
logger.info(
|
||||
"Here is the list of TEXT pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
return True
|
||||
except BaseException:
|
||||
logger.error(
|
||||
"Failed to get the list of TEXT pretrained models.")
|
||||
return False
|
||||
|
||||
elif self.task == 'tts':
|
||||
try:
|
||||
from ..tts.infer import pretrained_models
|
||||
logger.info(
|
||||
"Here is the list of TTS pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
return True
|
||||
except BaseException:
|
||||
logger.error("Failed to get the list of TTS pretrained models.")
|
||||
return False
|
||||
|
||||
@stats_wrapper
|
||||
def __call__(
|
||||
self,
|
||||
task: str=None, ):
|
||||
"""
|
||||
Python API to call an executor.
|
||||
"""
|
||||
self.task = task
|
||||
if self.task not in self.task_choices:
|
||||
print(
|
||||
"Please input correct speech task, choices = ['asr', 'cls', 'st', 'text', 'tts']"
|
||||
)
|
||||
|
||||
elif self.task == 'asr':
|
||||
try:
|
||||
from ..asr.infer import pretrained_models
|
||||
print(
|
||||
"Here is the list of ASR pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
except BaseException:
|
||||
print("Failed to get the list of ASR pretrained models.")
|
||||
|
||||
elif self.task == 'cls':
|
||||
try:
|
||||
from ..cls.infer import pretrained_models
|
||||
print(
|
||||
"Here is the list of CLS pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
except BaseException:
|
||||
print("Failed to get the list of CLS pretrained models.")
|
||||
|
||||
elif self.task == 'st':
|
||||
try:
|
||||
from ..st.infer import pretrained_models
|
||||
print(
|
||||
"Here is the list of ST pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
except BaseException:
|
||||
print("Failed to get the list of ST pretrained models.")
|
||||
|
||||
elif self.task == 'text':
|
||||
try:
|
||||
from ..text.infer import pretrained_models
|
||||
print(
|
||||
"Here is the list of TEXT pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
except BaseException:
|
||||
print("Failed to get the list of TEXT pretrained models.")
|
||||
|
||||
elif self.task == 'tts':
|
||||
try:
|
||||
from ..tts.infer import pretrained_models
|
||||
print(
|
||||
"Here is the list of TTS pretrained models released by PaddleSpeech that can be used by command line and python API"
|
||||
)
|
||||
self.show_support_models(pretrained_models)
|
||||
except BaseException:
|
||||
print("Failed to get the list of TTS pretrained models.")
|
||||
@ -1,48 +0,0 @@
|
||||
ConfigArgParse
|
||||
coverage
|
||||
editdistance
|
||||
g2p_en
|
||||
g2pM
|
||||
gpustat
|
||||
h5py
|
||||
inflect
|
||||
jieba
|
||||
jsonlines
|
||||
kaldiio
|
||||
librosa
|
||||
loguru
|
||||
matplotlib
|
||||
nara_wpe
|
||||
nltk
|
||||
paddleaudio
|
||||
paddlenlp
|
||||
paddlespeech_ctcdecoders
|
||||
paddlespeech_feat
|
||||
pandas
|
||||
phkit
|
||||
Pillow
|
||||
praatio==5.0.0
|
||||
pre-commit
|
||||
pybind11
|
||||
pypi-kenlm
|
||||
pypinyin
|
||||
python-dateutil
|
||||
pyworld
|
||||
resampy==0.2.2
|
||||
sacrebleu
|
||||
scipy
|
||||
sentencepiece~=0.1.96
|
||||
snakeviz
|
||||
soundfile~=0.10
|
||||
sox
|
||||
soxbindings
|
||||
textgrid
|
||||
timer
|
||||
tqdm
|
||||
typeguard
|
||||
unidecode
|
||||
visualdl
|
||||
webrtcvad
|
||||
yacs~=0.1.8
|
||||
yq
|
||||
zhon
|
||||
@ -0,0 +1,3 @@
|
||||
mkdir -p ./test_data
|
||||
wget -P ./test_data https://paddlespeech.bj.bcebos.com/datasets/unit_test/asr/static_ds2online_inputs.pickle
|
||||
python deepspeech2_online_model_test.py
|
||||
Loading…
Reference in new issue