Skip to content

D
DeepSpeech

PaddlePaddle / DeepSpeech
大约 2 年 前同步成功

通知 210
Star 8425
Fork 1598
D
DeepSpeech
提交 b7e908a5 编写于 作者: Y Yang Zhou
浏览文件
操作

merge audio

上级 d46cee6a f94cb483
展开全部 隐藏空白更改
内联 并排
Showing with 4316 addition and 59 deletion
audio/.gitignore 0 → 100644
浏览文件 @ b7e908a5
build
third_party/archives/
third_party/install/
audio/CMakeLists.txt
浏览文件 @ b7e908a5
cmake_minimum_required(VERSION 3.16 FATAL_ERROR)
cmake_minimum_required(VERSION 3.14 FATAL_ERROR)
project(paddleaudio VERSION 0.1) project(paddleaudio VERSION 0.1)
string(FIND "${CMAKE_CXX_FLAGS}" "-std=c++" env_cxx_standard)
# cmake dir # cmake dir
set(paddleaudio_cmake_dir ${PROJECT_SOURCE_DIR}/cmake) set(paddleaudio_cmake_dir ${PROJECT_SOURCE_DIR}/cmake)
...@@ -17,21 +18,26 @@ set(FETCHCONTENT_QUIET off) ...@@ -17,21 +18,26 @@ set(FETCHCONTENT_QUIET off)
get_filename_component(fc_patch "fc_patch" REALPATH BASE_DIR "${CMAKE_SOURCE_DIR}") get_filename_component(fc_patch "fc_patch" REALPATH BASE_DIR "${CMAKE_SOURCE_DIR}")
set(FETCHCONTENT_BASE_DIR ${fc_patch}) set(FETCHCONTENT_BASE_DIR ${fc_patch})
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} --std=c++14 -pthread -fPIC -O0 -Wall -g") set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -pthread -fPIC -O0 -Wall -g")
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_C_STANDARD 11)
set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
set(CMAKE_POSITION_INDEPENDENT_CODE ON)
option(BUILD_SOX "Build libsox statically" ON)
# see the thirdparty/kaldi/base/kaldi-types.h # checkout the thirdparty/kaldi/base/kaldi-types.h
# compile kaldi without openfst # compile kaldi without openfst
add_definitions("-DCOMPILE_WITHOUT_OPENFST") add_definitions("-DCOMPILE_WITHOUT_OPENFST")
include(openblas) #include(openblas)
include(pybind) include(pybind)
#set(fc_patch /paddle/mnt/zhouyang/wks2/PaddleSpeech/paddleaudio/fc_patch) set(fc_patch /paddle/mnt/zhouyang/wks2/PaddleSpeech/paddleaudio/fc_patch)
#set(OpenBLAS_INSTALL_PREFIX ${fc_patch}/openblas-install) set(OpenBLAS_INSTALL_PREFIX ${fc_patch}/openblas-install)
#link_directories(${OpenBLAS_INSTALL_PREFIX}/lib) link_directories(${OpenBLAS_INSTALL_PREFIX}/lib)
#include_directories(${OpenBLAS_INSTALL_PREFIX}/include) include_directories(${OpenBLAS_INSTALL_PREFIX}/include)
include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/third_party/kaldi) include_directories(${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}/third_party/kaldi)
#include_directories(${CMAKE_CURRENT_SOURCE_DIR}/third_party/pybind11/include)
include_directories(/usr/include/python3.7m) include_directories(/usr/include/python3.7m)
add_subdirectory(third_party) add_subdirectory(third_party)
add_subdirectory(kaldi_frontend) add_subdirectory(audio/csrc)
audio/audio/csrc/CMakeLists.txt 0 → 100644
浏览文件 @ b7e908a5
find_package(Python3 COMPONENTS Interpreter Development)
find_package(pybind11 CONFIG)
function(define_extension name sources libraries)
add_library(${name} SHARED ${sources})
target_include_directories(
${name} PRIVATE ${PROJECT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR} ${Python3_INCLUDE_DIRS} ${pybind11_INCLUDE_DIR})
target_link_libraries(${name} ${libraries})
set_target_properties(${name} PROPERTIES PREFIX "")
install(
TARGETS ${name}
LIBRARY DESTINATION .
)
endfunction()
if(BUILD_SOX)
set(
EXTENSION_SOURCES
pybind/pybind.cpp
pybind/sox/io.cpp
pybind/sox/utils.cpp
)
set(
LINK_LIBRARIES
libsox
)
define_extension(
_paddleaudio
"${EXTENSION_SOURCES}"
"${LINK_LIBRARIES}"
)
endif()
add_subdirectory(pybind/kaldi_frontend)
audio/audio/csrc/pybind/pybind.cpp 0 → 100644
浏览文件 @ b7e908a5
#include "pybind/sox/io.h"
PYBIND11_MODULE(_paddleaudio, m) {
m.def("get_info_file", &paddleaudio::sox_io::get_info_file,
"Get metadata of audio file.");
m.def("get_info_fileobj", &paddleaudio::sox_io::get_info_fileobj,
"Get metadata of audio in file object.");
}
\ No newline at end of file
audio/audio/csrc/pybind/sox/io.cpp 0 → 100644
浏览文件 @ b7e908a5
#include "pybind/sox/io.h"
#include "pybind/sox/utils.h"
using namespace paddleaudio::sox_utils;
namespace paddleaudio {
namespace sox_io {
auto get_info_file(const std::string &path, const std::string &format)
-> std::tuple<int64_t, int64_t, int64_t, int64_t, std::string> {
SoxFormat sf(sox_open_read(path.data(),
/*signal=*/nullptr,
/*encoding=*/nullptr,
/*filetype=*/format.empty() ? nullptr : format.data()));
validate_input_file(sf, path);
return std::make_tuple(
static_cast<int64_t>(sf->signal.rate),
static_cast<int64_t>(sf->signal.length / sf->signal.channels),
static_cast<int64_t>(sf->signal.channels),
static_cast<int64_t>(sf->encoding.bits_per_sample),
get_encoding(sf->encoding.encoding));
}
auto get_info_fileobj(py::object fileobj, const std::string &format)
-> std::tuple<int64_t, int64_t, int64_t, int64_t, std::string> {
const auto capacity = [&]() {
const auto bufsiz = get_buffer_size();
const int64_t kDefaultCapacityInBytes = 4096;
return (bufsiz > kDefaultCapacityInBytes) ? bufsiz
: kDefaultCapacityInBytes;
}();
std::string buffer(capacity, '\0');
auto *buf = const_cast<char *>(buffer.data());
auto num_read = read_fileobj(&fileobj, capacity, buf);
// If the file is shorter than 256, then libsox cannot read the header.
auto buf_size = (num_read > 256) ? num_read : 256;
SoxFormat sf(sox_open_mem_read(buf, buf_size,
/*signal=*/nullptr,
/*encoding=*/nullptr,
/*filetype=*/format.empty() ? nullptr : format.data()));
// In case of streamed data, length can be 0
validate_input_memfile(sf);
return std::make_tuple(
static_cast<int64_t>(sf->signal.rate),
static_cast<int64_t>(sf->signal.length / sf->signal.channels),
static_cast<int64_t>(sf->signal.channels),
static_cast<int64_t>(sf->encoding.bits_per_sample),
get_encoding(sf->encoding.encoding));
}
} // namespace paddleaudio
} // namespace sox_io
audio/audio/csrc/pybind/sox/io.h 0 → 100644
浏览文件 @ b7e908a5
#ifndef PADDLEAUDIO_PYBIND_SOX_IO_H
#define PADDLEAUDIO_PYBIND_SOX_IO_H
#include "pybind/sox/utils.h"
namespace paddleaudio {
namespace sox_io {
auto get_info_file(const std::string &path, const std::string &format)
-> std::tuple<int64_t, int64_t, int64_t, int64_t, std::string>;
auto get_info_fileobj(py::object fileobj, const std::string &format)
-> std::tuple<int64_t, int64_t, int64_t, int64_t, std::string>;
} // namespace paddleaudio
} // namespace sox_io
#endif
audio/audio/csrc/pybind/sox/utils.cpp 0 → 100644
浏览文件 @ b7e908a5
#include "pybind/sox/utils.h"
#include <sstream>
namespace paddleaudio {
namespace sox_utils {
SoxFormat::SoxFormat(sox_format_t *fd) noexcept : fd_(fd) {}
SoxFormat::~SoxFormat() { close(); }
sox_format_t *SoxFormat::operator->() const noexcept { return fd_; }
SoxFormat::operator sox_format_t *() const noexcept { return fd_; }
void SoxFormat::close() {
if (fd_ != nullptr) {
sox_close(fd_);
fd_ = nullptr;
}
}
auto read_fileobj(py::object *fileobj, const uint64_t size, char *buffer)
-> uint64_t {
uint64_t num_read = 0;
while (num_read < size) {
auto request = size - num_read;
auto chunk = static_cast<std::string>(
static_cast<py::bytes>(fileobj->attr("read")(request)));
auto chunk_len = chunk.length();
if (chunk_len == 0) {
break;
}
if (chunk_len > request) {
std::ostringstream message;
message << "Requested up to " << request << " bytes but, "
<< "received " << chunk_len << " bytes. "
<< "The given object does not confirm to read protocol of file "
"object.";
throw std::runtime_error(message.str());
}
memcpy(buffer, chunk.data(), chunk_len);
buffer += chunk_len;
num_read += chunk_len;
}
return num_read;
}
int64_t get_buffer_size() { return sox_get_globals()->bufsiz; }
void validate_input_file(const SoxFormat &sf, const std::string &path) {
if (static_cast<sox_format_t *>(sf) == nullptr) {
throw std::runtime_error("Error loading audio file: failed to open file " +
path);
}
if (sf->encoding.encoding == SOX_ENCODING_UNKNOWN) {
throw std::runtime_error("Error loading audio file: unknown encoding.");
}
}
void validate_input_memfile(const SoxFormat &sf) {
return validate_input_file(sf, "<in memory buffer>");
}
std::string get_encoding(sox_encoding_t encoding) {
switch (encoding) {
case SOX_ENCODING_UNKNOWN:
return "UNKNOWN";
case SOX_ENCODING_SIGN2:
return "PCM_S";
case SOX_ENCODING_UNSIGNED:
return "PCM_U";
case SOX_ENCODING_FLOAT:
return "PCM_F";
case SOX_ENCODING_FLAC:
return "FLAC";
case SOX_ENCODING_ULAW:
return "ULAW";
case SOX_ENCODING_ALAW:
return "ALAW";
case SOX_ENCODING_MP3:
return "MP3";
case SOX_ENCODING_VORBIS:
return "VORBIS";
case SOX_ENCODING_AMR_WB:
return "AMR_WB";
case SOX_ENCODING_AMR_NB:
return "AMR_NB";
case SOX_ENCODING_OPUS:
return "OPUS";
case SOX_ENCODING_GSM:
return "GSM";
default:
return "UNKNOWN";
}
}
} // namespace paddleaudio
} // namespace sox_utils
audio/audio/csrc/pybind/sox/utils.h 0 → 100644
浏览文件 @ b7e908a5
#ifndef PADDLEAUDIO_PYBIND_SOX_UTILS_H
#define PADDLEAUDIO_PYBIND_SOX_UTILS_H
#include <pybind11/pybind11.h>
#include <sox.h>
namespace py = pybind11;
namespace paddleaudio {
namespace sox_utils {
/// helper class to automatically close sox_format_t*
struct SoxFormat {
explicit SoxFormat(sox_format_t *fd) noexcept;
SoxFormat(const SoxFormat &other) = delete;
SoxFormat(SoxFormat &&other) = delete;
SoxFormat &operator=(const SoxFormat &other) = delete;
SoxFormat &operator=(SoxFormat &&other) = delete;
~SoxFormat();
sox_format_t *operator->() const noexcept;
operator sox_format_t *() const noexcept;
void close();
private:
sox_format_t *fd_;
};
auto read_fileobj(py::object *fileobj, uint64_t size, char *buffer) -> uint64_t;
int64_t get_buffer_size();
void validate_input_file(const SoxFormat &sf, const std::string &path);
void validate_input_memfile(const SoxFormat &sf);
std::string get_encoding(sox_encoding_t encoding);
} // namespace paddleaudio
} // namespace sox_utils
#endif
audio/third_party/CMakeLists.txt
浏览文件 @ b7e908a5
cmake_minimum_required(VERSION 3.14 FATAL_ERROR) set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fvisibility=hidden")
add_library(libsox INTERFACE)
if (BUILD_SOX)
add_subdirectory(sox)
target_include_directories(libsox INTERFACE ${SOX_INCLUDE_DIR})
target_link_libraries(libsox INTERFACE ${SOX_LIBRARIES})
endif()
include_directories( include_directories(
${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_CURRENT_SOURCE_DIR}
......
audio/third_party/kaldi/base 0 → 120000
浏览文件 @ b7e908a5
../../../speechx/speechx/kaldi/base
\ No newline at end of file
audio/third_party/kaldi/feat 0 → 120000
浏览文件 @ b7e908a5
../../../speechx/speechx/kaldi/feat
\ No newline at end of file
audio/third_party/kaldi/matrix 0 → 120000
浏览文件 @ b7e908a5
../../../speechx/speechx/kaldi/matrix
\ No newline at end of file
audio/third_party/kaldi/util 0 → 120000
浏览文件 @ b7e908a5
../../../speechx/speechx/kaldi/util
\ No newline at end of file
audio/third_party/patches/config.guess 0 → 100644
浏览文件 @ b7e908a5
此差异已折叠。
audio/third_party/patches/config.sub 0 → 100644
浏览文件 @ b7e908a5
此差异已折叠。
audio/third_party/patches/libmad.patch 0 → 100644
浏览文件 @ b7e908a5
See the followings for the origin of this patch
http://www.linuxfromscratch.org/blfs/view/svn/multimedia/libmad.html
http://www.linuxfromscratch.org/patches/blfs/svn/libmad-0.15.1b-fixes-1.patch
--- src/libmad/configure 2004-02-05 09:34:07.000000000 +0000
+++ src/libmad/configure.new 2020-06-30 21:10:28.528018931 +0000
@@ -19083,71 +19083,7 @@
if test "$GCC" = yes
then
- if test -z "$arch"
- then
- case "$host" in
- i386-*) ;;
- i?86-*) arch="-march=i486" ;;
- arm*-empeg-*) arch="-march=armv4 -mtune=strongarm1100" ;;
- armv4*-*) arch="-march=armv4 -mtune=strongarm" ;;
- powerpc-*) ;;
- mips*-agenda-*) arch="-mcpu=vr4100" ;;
- mips*-luxsonor-*) arch="-mips1 -mcpu=r3000 -Wa,-m4010" ;;
- esac
- fi
-
- case "$optimize" in
- -O|"-O "*)
- optimize="-O"
- optimize="$optimize -fforce-mem"
- optimize="$optimize -fforce-addr"
- : #x optimize="$optimize -finline-functions"
- : #- optimize="$optimize -fstrength-reduce"
- optimize="$optimize -fthread-jumps"
- optimize="$optimize -fcse-follow-jumps"
- optimize="$optimize -fcse-skip-blocks"
- : #x optimize="$optimize -frerun-cse-after-loop"
- : #x optimize="$optimize -frerun-loop-opt"
- : #x optimize="$optimize -fgcse"
- optimize="$optimize -fexpensive-optimizations"
- optimize="$optimize -fregmove"
- : #* optimize="$optimize -fdelayed-branch"
- : #x optimize="$optimize -fschedule-insns"
- optimize="$optimize -fschedule-insns2"
- : #? optimize="$optimize -ffunction-sections"
- : #? optimize="$optimize -fcaller-saves"
- : #> optimize="$optimize -funroll-loops"
- : #> optimize="$optimize -funroll-all-loops"
- : #x optimize="$optimize -fmove-all-movables"
- : #x optimize="$optimize -freduce-all-givs"
- : #? optimize="$optimize -fstrict-aliasing"
- : #* optimize="$optimize -fstructure-noalias"
-
- case "$host" in
- arm*-*)
- optimize="$optimize -fstrength-reduce"
- ;;
- mips*-*)
- optimize="$optimize -fstrength-reduce"
- optimize="$optimize -finline-functions"
- ;;
- i?86-*)
- optimize="$optimize -fstrength-reduce"
- ;;
- powerpc-apple-*)
- # this triggers an internal compiler error with gcc2
- : #optimize="$optimize -fstrength-reduce"
-
- # this is really only beneficial with gcc3
- : #optimize="$optimize -finline-functions"
- ;;
- *)
- # this sometimes provokes bugs in gcc 2.95.2
- : #optimize="$optimize -fstrength-reduce"
- ;;
- esac
- ;;
- esac
+ optimize="-O2"
fi
case "$host" in
@@ -21497,6 +21433,7 @@
then
case "$host" in
i?86-*) FPM="INTEL" ;;
+ x86_64*) FPM="64BIT" ;;
arm*-*) FPM="ARM" ;;
mips*-*) FPM="MIPS" ;;
sparc*-*) FPM="SPARC" ;;
audio/third_party/patches/sox.patch 0 → 100644
浏览文件 @ b7e908a5
See https://github.com/pytorch/audio/pull/1297
diff -ru sox/src/formats.c sox/src/formats.c
--- sox/src/formats.c 2014-10-26 19:55:50.000000000 -0700
+++ sox/src/formats.c 2021-02-22 16:01:02.833144070 -0800
@@ -333,6 +333,10 @@
assert(ft);
if (!ft->fp)
return sox_false;
- fstat(fileno((FILE*)ft->fp), &st);
+ int fd = fileno((FILE*)ft->fp);
+ if (fd < 0)
+ return sox_false;
+ if (fstat(fd, &st) < 0)
+ return sox_false;
return ((st.st_mode & S_IFMT) == S_IFREG);
}
audio/third_party/sox/CMakeLists.txt 0 → 100644
浏览文件 @ b7e908a5
find_package(PkgConfig REQUIRED)
include(ExternalProject)
set(INSTALL_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../install)
set(ARCHIVE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../archives)
set(patch_dir ${PROJECT_SOURCE_DIR}/third_party/patches)
set(COMMON_ARGS --quiet --disable-shared --enable-static --prefix=${INSTALL_DIR} --with-pic --disable-dependency-tracking --disable-debug --disable-examples --disable-doc)
# To pass custom environment variables to ExternalProject_Add command,
# we need to do `${CMAKE_COMMAND} -E env ${envs} <COMMANAD>`.
# https://stackoverflow.com/a/62437353
# We constrcut the custom environment variables here
set(envs
"PKG_CONFIG_PATH=${INSTALL_DIR}/lib/pkgconfig"
"LDFLAGS=-L${INSTALL_DIR}/lib $ENV{LDFLAGS}"
"CFLAGS=-I${INSTALL_DIR}/include -fvisibility=hidden $ENV{CFLAGS}"
)
ExternalProject_Add(mad
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://downloads.sourceforge.net/project/mad/libmad/0.15.1b/libmad-0.15.1b.tar.gz
URL_HASH SHA256=bbfac3ed6bfbc2823d3775ebb931087371e142bb0e9bb1bee51a76a6e0078690
PATCH_COMMAND patch < ${patch_dir}/libmad.patch && cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/mad/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/mad/configure ${COMMON_ARGS}
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(amr
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://sourceforge.net/projects/opencore-amr/files/opencore-amr/opencore-amr-0.1.5.tar.gz
URL_HASH SHA256=2c006cb9d5f651bfb5e60156dbff6af3c9d35c7bbcc9015308c0aff1e14cd341
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/amr/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/amr/configure ${COMMON_ARGS}
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(lame
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://downloads.sourceforge.net/project/lame/lame/3.99/lame-3.99.5.tar.gz
URL_HASH SHA256=24346b4158e4af3bd9f2e194bb23eb473c75fb7377011523353196b19b9a23ff
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/lame/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/lame/configure ${COMMON_ARGS} --enable-nasm
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(ogg
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://ftp.osuosl.org/pub/xiph/releases/ogg/libogg-1.3.3.tar.gz
URL_HASH SHA256=c2e8a485110b97550f453226ec644ebac6cb29d1caef2902c007edab4308d985
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/ogg/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/ogg/configure ${COMMON_ARGS}
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(flac
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DEPENDS ogg
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://ftp.osuosl.org/pub/xiph/releases/flac/flac-1.3.2.tar.xz
URL_HASH SHA256=91cfc3ed61dc40f47f050a109b08610667d73477af6ef36dcad31c31a4a8d53f
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/flac/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/flac/configure ${COMMON_ARGS} --with-ogg --disable-cpplibs
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(vorbis
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DEPENDS ogg
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://ftp.osuosl.org/pub/xiph/releases/vorbis/libvorbis-1.3.6.tar.gz
URL_HASH SHA256=6ed40e0241089a42c48604dc00e362beee00036af2d8b3f46338031c9e0351cb
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/vorbis/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/vorbis/configure ${COMMON_ARGS} --with-ogg
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(opus
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DEPENDS ogg
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://ftp.osuosl.org/pub/xiph/releases/opus/opus-1.3.1.tar.gz
URL_HASH SHA256=65b58e1e25b2a114157014736a3d9dfeaad8d41be1c8179866f144a2fb44ff9d
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/opus/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/opus/configure ${COMMON_ARGS} --with-ogg
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
ExternalProject_Add(opusfile
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DEPENDS opus
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://ftp.osuosl.org/pub/xiph/releases/opus/opusfile-0.12.tar.gz
URL_HASH SHA256=118d8601c12dd6a44f52423e68ca9083cc9f2bfe72da7a8c1acb22a80ae3550b
PATCH_COMMAND cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/opusfile/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/opusfile/configure ${COMMON_ARGS} --disable-http
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
set(SOX_OPTIONS
--disable-openmp
--with-amrnb
--with-amrwb
--with-flac
--with-lame
--with-mad
--with-oggvorbis
--with-opus
--without-alsa
--without-ao
--without-coreaudio
--without-oss
--without-id3tag
--without-ladspa
--without-magic
--without-png
--without-pulseaudio
--without-sndfile
--without-sndio
--without-sunaudio
--without-waveaudio
--without-wavpack
--without-twolame
)
set(SOX_LIBRARIES
${INSTALL_DIR}/lib/libsox.a
${INSTALL_DIR}/lib/libopencore-amrnb.a
${INSTALL_DIR}/lib/libopencore-amrwb.a
${INSTALL_DIR}/lib/libmad.a
${INSTALL_DIR}/lib/libmp3lame.a
${INSTALL_DIR}/lib/libFLAC.a
${INSTALL_DIR}/lib/libopusfile.a
${INSTALL_DIR}/lib/libopus.a
${INSTALL_DIR}/lib/libvorbisenc.a
${INSTALL_DIR}/lib/libvorbisfile.a
${INSTALL_DIR}/lib/libvorbis.a
${INSTALL_DIR}/lib/libogg.a
)
ExternalProject_Add(sox
PREFIX ${CMAKE_CURRENT_BINARY_DIR}
DEPENDS ogg flac vorbis opusfile lame mad amr
DOWNLOAD_DIR ${ARCHIVE_DIR}
URL https://downloads.sourceforge.net/project/sox/sox/14.4.2/sox-14.4.2.tar.bz2
URL_HASH SHA256=81a6956d4330e75b5827316e44ae381e6f1e8928003c6aa45896da9041ea149c
PATCH_COMMAND patch -p1 < ${patch_dir}/sox.patch && cp ${patch_dir}/config.guess ${patch_dir}/config.sub ${CMAKE_CURRENT_BINARY_DIR}/src/sox/
CONFIGURE_COMMAND ${CMAKE_COMMAND} -E env ${envs} ${CMAKE_CURRENT_BINARY_DIR}/src/sox/configure ${COMMON_ARGS} ${SOX_OPTIONS}
BUILD_BYPRODUCTS ${SOX_LIBRARIES}
DOWNLOAD_NO_PROGRESS ON
LOG_DOWNLOAD ON
LOG_UPDATE ON
LOG_CONFIGURE ON
LOG_BUILD ON
LOG_INSTALL ON
LOG_MERGED_STDOUTERR ON
LOG_OUTPUT_ON_FAILURE ON
)
add_dependencies(libsox sox)
set(SOX_INCLUDE_DIR ${INSTALL_DIR}/include PARENT_SCOPE)
set(SOX_LIBRARIES ${SOX_LIBRARIES} PARENT_SCOPE)
docs/source/install.md
浏览文件 @ b7e908a5
...@@ -139,28 +139,13 @@ pip install . -i https://pypi.tuna.tsinghua.edu.cn/simple ...@@ -139,28 +139,13 @@ pip install . -i https://pypi.tuna.tsinghua.edu.cn/simple
To avoid the trouble of environment setup, running in a Docker container is highly recommended. Otherwise, if you work on `Ubuntu` with `root` privilege, you can still complete the installation. To avoid the trouble of environment setup, running in a Docker container is highly recommended. Otherwise, if you work on `Ubuntu` with `root` privilege, you can still complete the installation.
### Choice 1: Running in Docker Container (Recommend) ### Choice 1: Running in Docker Container (Recommend)
Docker is an open-source tool to build, ship, and run distributed applications in an isolated environment. A Docker image for this project has been provided in [hub.docker.com](https://hub.docker.com) with dependencies of cuda and cudnn installed. This Docker image requires the support of NVIDIA GPU, so please make sure its availability and the [nvidia-docker](https://github.com/NVIDIA/nvidia-docker) has been installed. Docker is an open-source tool to build, ship, and run distributed applications in an isolated environment. If you do not have a Docker environment, please refer to [Docker](https://www.docker.com/). If you will use GPU version, you also need to install [nvidia-docker](https://github.com/NVIDIA/nvidia-docker).
Take several steps to launch the Docker image: We provide docker images containing the latest PaddleSpeech code, and all environment and package dependencies are pre-installed. All you have to do is to **pull and run the docker image**. Then you can enjoy PaddleSpeech without any extra steps.
- Download the Docker image
For example, pull paddle 2.2.0 image: Get these images and guidance in [docker hub](https://hub.docker.com/repository/docker/paddlecloud/paddlespeech), including CPU, GPU, ROCm environment versions.
```bash
sudo nvidia-docker pull registry.baidubce.com/paddlepaddle/paddle:2.2.0-gpu-cuda10.2-cudnn7 If you have some customized requirements about automatic building docker images, you can get it in github repo [PaddlePaddle/PaddleCloud](https://github.com/PaddlePaddle/PaddleCloud/tree/main/tekton).
```
- Clone this repository
```bash
git clone https://github.com/PaddlePaddle/PaddleSpeech.git
```
- Run the Docker image
```bash
sudo nvidia-docker run --net=host --ipc=host --rm -it -v $(pwd)/PaddleSpeech:/PaddleSpeech registry.baidubce.com/paddlepaddle/paddle:2.2.0-gpu-cuda10.2-cudnn7 /bin/bash
```
- Enter PaddleSpeech directory.
```bash
cd /PaddleSpeech
```
Now you can execute training, inference, and hyper-parameters tuning in Docker container.
### Choice 2: Running in Ubuntu with Root Privilege ### Choice 2: Running in Ubuntu with Root Privilege
- Install `build-essential` by apt - Install `build-essential` by apt
......
docs/source/install_cn.md
浏览文件 @ b7e908a5
...@@ -130,26 +130,14 @@ pip install . -i https://pypi.tuna.tsinghua.edu.cn/simple ...@@ -130,26 +130,14 @@ pip install . -i https://pypi.tuna.tsinghua.edu.cn/simple
- 选择 2： 使用`Ubuntu` ，并且拥有 root 权限。 - 选择 2： 使用`Ubuntu` ，并且拥有 root 权限。
为了避免各种环境配置问题，我们非常推荐你使用 docker 容器。如果你不想使用 docker，但是可以使用拥有 root 权限的 Ubuntu 系统，你也可以完成**困难**方式的安装。 为了避免各种环境配置问题，我们非常推荐你使用 docker 容器。如果你不想使用 docker，但是可以使用拥有 root 权限的 Ubuntu 系统，你也可以完成**困难**方式的安装。
### 选择1： 使用Docker容器（推荐） ### 选择1： 使用 Docker 容器（推荐）
Docker 是一种开源工具，用于在和系统本身环境相隔离的环境中构建、发布和运行各类应用程序。你可以访问 [hub.docker.com](https://hub.docker.com) 来下载各种版本的 docker，目前已经有适用于 `PaddleSpeech` 的 docker 提供在了该网站上。Docker 镜像需要使用 Nvidia GPU，所以你也需要提前安装好 [nvidia-docker](https://github.com/NVIDIA/nvidia-docker) 。 Docker 是一种开源工具，用于在和系统本身环境相隔离的环境中构建、发布和运行各类应用程序。如果您没有 Docker 运行环境，请参考 [Docker 官网](https://www.docker.com/)进行安装，如果您准备使用 GPU 版本镜像，还需要提前安装好 [nvidia-docker](https://github.com/NVIDIA/nvidia-docker) 。
你需要完成几个步骤来启动docker：
- 下载 docker 镜像: 我们提供了包含最新 PaddleSpeech 代码的 docker 镜像，并预先安装好了所有的环境和库依赖，您只需要**拉取并运行 docker 镜像**，无需其他任何额外操作，即可开始享用 PaddleSpeech 的所有功能。
例如，拉取 paddle2.2.0 镜像：
```bash 在 [Docker Hub](https://hub.docker.com/repository/docker/paddlecloud/paddlespeech) 中获取这些镜像及相应的使用指南，包括 CPU、GPU、ROCm 版本。
sudo nvidia-docker pull registry.baidubce.com/paddlepaddle/paddle:2.2.0-gpu-cuda10.2-cudnn7
``` 如果您对自动化制作docker镜像感兴趣，或有自定义需求，请访问 [PaddlePaddle/PaddleCloud](https://github.com/PaddlePaddle/PaddleCloud/tree/main/tekton) 做进一步了解。
- 克隆 `PaddleSpeech` 仓库
```bash
git clone https://github.com/PaddlePaddle/PaddleSpeech.git
```
- 启动 docker 镜像
```bash
sudo nvidia-docker run --net=host --ipc=host --rm -it -v $(pwd)/PaddleSpeech:/PaddleSpeech registry.baidubce.com/paddlepaddle/paddle:2.2.0-gpu-cuda10.2-cudnn7 /bin/bash
```
- 进入 PaddleSpeech 目录
```bash
cd /PaddleSpeech
```
完成这些以后，你就可以在 docker 容器中执行训练、推理和超参 fine-tune。 完成这些以后，你就可以在 docker 容器中执行训练、推理和超参 fine-tune。
### 选择2： 使用有 root 权限的 Ubuntu ### 选择2： 使用有 root 权限的 Ubuntu
- 使用apt安装 `build-essential` - 使用apt安装 `build-essential`
......
setup.py
浏览文件 @ b7e908a5
...@@ -32,13 +32,43 @@ VERSION = '0.0.0' ...@@ -32,13 +32,43 @@ VERSION = '0.0.0'
COMMITID = 'none' COMMITID = 'none'
base = [ base = [
"editdistance", "g2p_en", "g2pM", "h5py", "inflect", "jieba", "jsonlines", "editdistance",
"kaldiio", "librosa==0.8.1", "loguru", "matplotlib", "nara_wpe", "g2p_en",
"onnxruntime", "pandas", "paddlenlp", "paddlespeech_feat", "praatio==5.0.0", "g2pM",
"pypinyin", "pypinyin-dict", "python-dateutil", "pyworld", "resampy==0.2.2", "h5py",
"sacrebleu", "scipy", "sentencepiece~=0.1.96", "soundfile~=0.10", "inflect",
"textgrid", "timer", "tqdm", "typeguard", "visualdl", "webrtcvad", "jieba",
"yacs~=0.1.8", "prettytable", "zhon", 'colorlog', 'pathos == 0.2.8' "jsonlines",
"kaldiio",
"librosa==0.8.1",
"loguru",
"matplotlib",
"nara_wpe",
"onnxruntime",
"pandas",
"paddlenlp",
"paddlespeech_feat",
"praatio==5.0.0",
"pypinyin",
"pypinyin-dict",
"python-dateutil",
"pyworld",
"resampy==0.2.2",
"sacrebleu",
"scipy",
"sentencepiece~=0.1.96",
"soundfile~=0.10",
"textgrid",
"timer",
"tqdm",
"typeguard",
"visualdl",
"webrtcvad",
"yacs~=0.1.8",
"prettytable",
"zhon",
'colorlog',
'pathos == 0.2.8'
] ]
server = [ server = [
...@@ -239,6 +269,9 @@ setup_info = dict( ...@@ -239,6 +269,9 @@ setup_info = dict(
"speech", "speech",
"asr", "asr",
"tts", "tts",
"streaming asr"
"streaming tts"
"audio process"
"speaker verfication", "speaker verfication",
"speech classfication", "speech classfication",
"text frontend", "text frontend",
...@@ -252,7 +285,11 @@ setup_info = dict( ...@@ -252,7 +285,11 @@ setup_info = dict(
"fastspeech", "fastspeech",
"vocoder", "vocoder",
"pwgan", "pwgan",
"melgan",
"mb-melgan",
"hifigan",
"gan", "gan",
"wfst decoder",
], ],
python_requires='>=3.7', python_requires='>=3.7',
install_requires=requirements["install"], install_requires=requirements["install"],
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或