mirror/cube-studio
2
0
Fork 0
mirror of https://github.com/tencentmusic/cube-studio.git synced 2025-02-17 14:40:28 +08:00
Code Issues Packages Projects Releases Wiki Activity

include paddle-speech

Browse Source 
UPDATE paddle-speech
This commit is contained in:
JLWL 2022-11-16 15:16:27 +08:00
parent f8dc37ae34
commit 43282761d9
15 changed files with 30 additions and 160 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
aihub/deep-learning/paddlespeech-asr/Dockerfile
View File

@ -1,5 +1,5 @@
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech .
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech-asr .
FROM ccr.ccs.tencentyun.com/cube-studio/aihub:base-python3.8
# 安装基础环境
WORKDIR /

4
aihub/deep-learning/paddlespeech-asr/app.py
View File

@ -25,13 +25,13 @@ class Speech_Asr_Model(Model):
inference_inputs = [
Field(type=Field_type.audio, name='voice_file_path', label='语音文件',
describe='可支持上传一个语音文件'),
describe='上传一个语音文件试试吧~'),
]
web_examples = [
{
"label": "示例1",
"input": {
"voice_file_path": "a.wav"
"voice_file_path": "zh.wav"
}
}
]

3
aihub/deep-learning/paddlespeech-asr/init.sh
View File

@ -4,3 +4,6 @@ conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
pip install paddlespeech -i https://pypi.tuna.tsinghua.edu.cn/simple
cd /app/
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav

2
aihub/deep-learning/paddlespeech-cls/Dockerfile
View File

@ -1,5 +1,5 @@
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech .
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech-cls .
FROM ccr.ccs.tencentyun.com/cube-studio/aihub:base-python3.8
# 安装基础环境
WORKDIR /

4
aihub/deep-learning/paddlespeech-cls/app.py
View File

@ -25,13 +25,13 @@ class Speech_Cls_Model(Model):
inference_inputs = [
Field(type=Field_type.audio, name='voice_file_path', label='语音文件',
describe='上传一个语音文件吧~'),
describe='上传一个语音文件试试吧~'),
]
web_examples = [
{
"label": "示例1",
"input": {
"voice_file_path": "a.wav"
"voice_file_path": "zh.wav"
}
}
]

3
aihub/deep-learning/paddlespeech-cls/init.sh
View File

@ -4,3 +4,6 @@ conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
pip install paddlespeech -i https://pypi.tuna.tsinghua.edu.cn/simple
cd /app/
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav

2
aihub/deep-learning/paddlespeech-st/Dockerfile
View File

@ -1,5 +1,5 @@
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech .
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech-st .
FROM ccr.ccs.tencentyun.com/cube-studio/aihub:base-python3.8
# 安装基础环境
WORKDIR /

2
aihub/deep-learning/paddlespeech-st/app.py
View File

@ -31,7 +31,7 @@ class Speech_St_Model(Model):
{
"label": "示例1",
"input": {
"voice_file_path": "a.wav"
"voice_file_path": "en.wav"
}
}
]

3
aihub/deep-learning/paddlespeech-st/init.sh
View File

@ -4,3 +4,6 @@ conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
pip install paddlespeech -i https://pypi.tuna.tsinghua.edu.cn/simple
cd /app/
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav

2
aihub/deep-learning/paddlespeech-tts/Dockerfile
View File

@ -1,5 +1,5 @@
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech .
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech-tts .
FROM ccr.ccs.tencentyun.com/cube-studio/aihub:base-python3.8
# 安装基础环境
WORKDIR /

22
aihub/deep-learning/paddlespeech-tts/app.py
View File

@ -32,15 +32,18 @@ class Speech_Tts_Model(Model):
inference_inputs = [
Field(type=Field_type.text, name='text', label='语音转文本',
describe='输入文本', default='cube studio 是个云原生一站式机器学习平台，欢迎大家体验！'),
Field(type=Field_type.text, name='spk_id', label='说话人ID',
describe='0-283可选，不一样的ID会带来不一样的声音', default=0),
# Field(type=Field_type.text, name='spk_id', label='说话人ID',
# describe='0-283可选，不一样的ID会带来不一样的声音', default=0),
Field(type=Field_type.text_select, name='spk_id', label='说话人', default='0',
choices=['1', '2', '3', '4', '5'])
]
web_examples = [
{
"label": "示例1",
"input": {
"text": "今天天气不错",
"spk_id": 0,
"spk_id": '0',
}
}
]
@ -57,19 +60,12 @@ class Speech_Tts_Model(Model):
@pysnooper.snoop()
def inference(self, text, spk_id=0):
tts = self.tts
file_name = f"{datetime.datetime.now().strftime('%Y%m%d%H%M%S')}-output.wav"
os.makedirs('result', exist_ok=True)
if spk_id:
if str(spk_id).isdigit():
if 0 <= int(spk_id) <= 283:
spk_id = int(spk_id)
else:
back = [{'text': '警告：数字需要在0~283之间哦~'}]
return back
else:
back = [{'text': '警告：说话人仅支持数字哦~'}]
return back
spk_id = int(spk_id) - 1 if int(spk_id) > 0 else int(spk_id)
else:
spk_id = 0
file_name = f"result/{datetime.datetime.now().strftime('%Y%m%d%H%M%S')}-output.wav"
tts(text=text, output=file_name, am='fastspeech2_mix', voc='hifigan_csmsc', lang='mix', spk_id=spk_id)
back = [
{

3
aihub/deep-learning/paddlespeech-tts/init.sh
View File

@ -4,3 +4,6 @@ conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
pip install paddlespeech -i https://pypi.tuna.tsinghua.edu.cn/simple
cd /app/
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav

12
aihub/deep-learning/paddlespeech/Dockerfile
View File

@ -1,12 +0,0 @@
# 构建应用镜像 docker build -t ccr.ccs.tencentyun.com/cube-studio/aihub:paddlespeech .
FROM ccr.ccs.tencentyun.com/cube-studio/aihub:base-python3.8
# 安装基础环境
WORKDIR /
COPY init.sh /init.sh
RUN bash /init.sh
# 安装文件
WORKDIR /app
COPY * /app/
ENTRYPOINT ["python", "app.py"]

118
aihub/deep-learning/paddlespeech/app-.py
View File

@ -1,118 +0,0 @@
import base64
import io, sys, os
from cubestudio.aihub.model import Model
from cubestudio.aihub.docker import Docker
from cubestudio.aihub.web.server import Server, Field, Field_type
import pysnooper
import os
import argparse, os, re
import datetime
import time
# speech所需内容，可根据需要注释无需的加载项
# 语音识别
from paddlespeech.cli.asr.infer import ASRExecutor
# 语音合成
from paddlespeech.cli.tts.infer import TTSExecutor
# 声音场景分类
from paddlespeech.cli.cls.infer import CLSExecutor
# 语音翻译
from paddlespeech.cli.st.infer import STExecutor
class Speech_Model(Model):
# 模型基础信息定义
name = 'paddle-speech'
label = '语音处理'
describe = "涵盖功能有语音转文字，文字转语音，语音翻译，语音场景识别"
field = "智能识别"
scenes = "语音处理"
status = 'online'
version = 'v20221114'
doc = 'https://github.com/PaddlePaddle/PaddleSpeech' # 'https://帮助文档的链接地址'
# pic = 'https://images.nightcafe.studio//assets/stable-tile.jpg' # https://应用描述的缩略图/可以直接使用应用内的图片文件地址
inference_resource = {
"resource_gpu": "1"
}
inference_inputs = [
Field(type=Field_type.text, name='text', label='语音合成文本',
describe='输入文本', default='cube studio 是个云原生一站式机器学习平台，欢迎大家体验！'),
Field(type=Field_type.audio, name='voice', label='语音文件',
describe='可支持上传一个语音文件，在下方选择一种想要进行的操作吧~'),
]
web_examples = [
{
"label": "语音合成",
"data": {
"work": ['语音分类', '语音识别', '...'],
"text": '今天天气不错',
"name": 'a.wav',
}
}
]
def __init__(self):
self.st = None
self.asr = None
self.tts = None
self.cls = None
# 加载模型
# @pysnooper.snoop()
def load_model(self):
self.cls = CLSExecutor() # 语音分类
self.tts = TTSExecutor() # 语音合成
self.asr = ASRExecutor() # 语音识别
self.st = STExecutor() # 语音翻译
# 推理
@pysnooper.snoop()
def inference(self, **kwargs):
"""
kwargs例子：
{
"label": "语音合成",
"data": {
"work": ['语音分类', '语音识别', '...'],
"text": '今天天气不错',
"file_name": 'a.wav',
}
}
"""
result = ''
file_PATH = ''
if kwargs['label'] == '语音合成':
tts = self.tts
file_name = f"{kwargs['data']['name']}-{datetime.datetime.now().strftime('%Y%m%d%H%M%S')}-output.wav"
tts(text=kwargs['data']['text'], output=file_name)
file_PATH = file_name
elif kwargs['label'] == '语音内容识别':
# file = open(kwargs['data']['name'], "wb") # 写入二进制文件
# text = base64.b64decode(kwargs['data']['text']) # 进行解码
# file.write(text)
if '语音分类' in kwargs['data']['works']:
cls = self.cls
result += '语音分类结果： '
result += cls(audio_file=kwargs['data']['file_name']) + '\r\n'
if '语音识别' in kwargs['data']['works']:
asr = self.asr
result += '语音识别结果： '
result += asr(audio_file=kwargs['data']['file_name']) + '\r\n'
if '语音翻译' in kwargs['data']['works']:
st = self.st
result += '语音翻译结果： '
result += ''.join(st(audio_file=kwargs['data']['file_name'])) + '\r\n'
return file_PATH, result
model = Speech_Model()
# model.load_model()
# result = model.inference(prompt='a photograph of an astronaut riding a horse',device='cpu') # 测试
# print(result)
# 启动服务
server = Server(model=model)
server.server(port=8080)

8
aihub/deep-learning/paddlespeech/init.sh
View File

@ -1,8 +0,0 @@
apt install build-essential
git clone https://github.com/PaddlePaddle/PaddleSpeech.git
cd PaddleSpeech
conda install -y -c conda-forge sox libsndfile swig bzip2 libflac bc
pip install pytest-runner -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
pip install -e .