Add BUILD_MAD option and default to OFF (#2354)

Summary: libmad integration should be enabled only from source-build Pull Request resolved: https://github.com/pytorch/audio/pull/2354 Reviewed By: nateanl Differential Revision: D36012035 Pulled By: mthrok fbshipit-source-id: adeda8cbfd418f96245909cae6862b648a6915a7

Add BUILD_MAD option and default to OFF (#2354)
Summary: libmad integration should be enabled only from source-build Pull Request resolved: https://github.com/pytorch/audio/pull/2354 Reviewed By: nateanl Differential Revision: D36012035 Pulled By: mthrok fbshipit-source-id: adeda8cbfd418f96245909cae6862b648a6915a7
a71e3a40 · moto · Facebook GitHub Bot · 3cf7f264 · a71e3a40 · a71e3a40
Commit a71e3a40 authored Apr 28, 2022 by moto Committed by Facebook GitHub Bot Apr 28, 2022
8 changed files
--- a/.circleci/config.yml
+++ b/.circleci/config.yml
@@ -495,6 +495,7 @@ jobs:
          name: Install torchaudio
          command: .circleci/unittest/linux/scripts/install.sh
          environment:
+              BUILD_MAD: true
              BUILD_FFMPEG: true
      - run:
          name: Run tests
@@ -528,7 +529,7 @@ jobs:
          command: docker run -t --gpus all -e PYTHON_VERSION -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/setup_env.sh
      - run:
          name: Install torchaudio
-          command: docker run -t --gpus all -e UPLOAD_CHANNEL -e CONDA_CHANNEL_FLAGS -e BUILD_FFMPEG=1 -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/install.sh
+          command: docker run -t --gpus all -e UPLOAD_CHANNEL -e CONDA_CHANNEL_FLAGS -e BUILD_FFMPEG=1 -e BUILD_MAD=1 -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/install.sh
      - run:
          name: Run tests
          environment:
@@ -641,6 +642,7 @@ jobs:
          command: .circleci/unittest/linux/scripts/install.sh
          environment:
              BUILD_FFMPEG: true
+              BUILD_MAD: true
      - run:
          name: Run tests
          command: .circleci/unittest/linux/scripts/run_test.sh

--- a/.circleci/config.yml.in
+++ b/.circleci/config.yml.in
@@ -495,6 +495,7 @@ jobs:
          name: Install torchaudio
          command: .circleci/unittest/linux/scripts/install.sh
          environment:
+              BUILD_MAD: true
              BUILD_FFMPEG: true
      - run:
          name: Run tests
@@ -528,7 +529,7 @@ jobs:
          command: docker run -t --gpus all -e PYTHON_VERSION -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/setup_env.sh
      - run:
          name: Install torchaudio
-          command: docker run -t --gpus all -e UPLOAD_CHANNEL -e CONDA_CHANNEL_FLAGS -e BUILD_FFMPEG=1 -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/install.sh
+          command: docker run -t --gpus all -e UPLOAD_CHANNEL -e CONDA_CHANNEL_FLAGS -e BUILD_FFMPEG=1 -e BUILD_MAD=1 -v $PWD:$PWD -w $PWD "${image_name}" .circleci/unittest/linux/scripts/install.sh
      - run:
          name: Run tests
          environment:
@@ -641,6 +642,7 @@ jobs:
          command: .circleci/unittest/linux/scripts/install.sh
          environment:
              BUILD_FFMPEG: true
+              BUILD_MAD: true
      - run:
          name: Run tests
          command: .circleci/unittest/linux/scripts/run_test.sh

--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -57,6 +57,7 @@ endif()

 # Options
 option(BUILD_SOX "Build libsox statically" ON)
+option(BUILD_MAD "Enable libmad" OFF)
 option(BUILD_FFMPEG "Enable ffmpeg-based features" OFF)
 option(BUILD_KALDI "Build kaldi statically" ON)
 option(BUILD_RNNT "Enable RNN transducer" ON)

--- a/examples/tutorials/audio_data_augmentation_tutorial.py
+++ b/examples/tutorials/audio_data_augmentation_tutorial.py
@@ -411,7 +411,6 @@ plot_specgram(waveform, sample_rate, title="Original")
 configs = [
    ({"format": "wav", "encoding": "ULAW", "bits_per_sample": 8}, "8 bit mu-law"),
    ({"format": "gsm"}, "GSM-FR"),
-    ({"format": "mp3", "compression": -9}, "MP3"),
    ({"format": "vorbis", "compression": -1}, "Vorbis"),
 ]
 waveforms = []
@@ -441,19 +440,12 @@ play_audio(waveforms[0], sample_rate)

 play_audio(waveforms[1], sample_rate)

-######################################################################
-# MP3:
-# ~~~~
-#
-
-play_audio(waveforms[2], sample_rate)
-
 ######################################################################
 # Vorbis:
 # ~~~~~~~
 #

-play_audio(waveforms[3], sample_rate)
+play_audio(waveforms[2], sample_rate)

 ######################################################################
 # Simulating a phone recoding

--- a/examples/tutorials/audio_io_tutorial.py
+++ b/examples/tutorials/audio_io_tutorial.py
@@ -225,13 +225,16 @@ print(metadata)
 #    variable bit rate (such as MP3).
 # -  ``num_frames`` can be ``0`` for GSM-FR format.
 #
-
-metadata = torchaudio.info(SAMPLE_MP3_PATH)
-print(metadata)
-
-metadata = torchaudio.info(SAMPLE_GSM_PATH)
-print(metadata)
-
+# .. code::
+#
+#    metadata = torchaudio.info(SAMPLE_MP3_PATH)
+#    print(metadata)
+#
+#    metadata = torchaudio.info(SAMPLE_GSM_PATH)
+#    print(metadata)
+#
+#    >>> AudioMetaData(sample_rate=44100, num_frames=110559, num_channels=2, bits_per_sample=0, encoding=MP3)
+#    >>> AudioMetaData(sample_rate=8000, num_frames=0, num_channels=1, bits_per_sample=0, encoding=GSM)

 ######################################################################
 # Querying file-like object
@@ -256,13 +259,18 @@ print(metadata)
 # -  Use argument ``format`` to specify the audio format of the input.
 # -  The returned metadata has ``num_frames = 0``
 #
-
-print("Source:", SAMPLE_MP3_URL)
-with requests.get(SAMPLE_MP3_URL, stream=True) as response:
-    metadata = torchaudio.info(response.raw, format="mp3")
-
-    print(f"Fetched {response.raw.tell()} bytes.")
-print(metadata)
+# .. code::
+#
+#    print("Source:", SAMPLE_MP3_URL)
+#    with requests.get(SAMPLE_MP3_URL, stream=True) as response:
+#        metadata = torchaudio.info(response.raw, format="mp3")
+#
+#        print(f"Fetched {response.raw.tell()} bytes.")
+#    print(metadata)
+#
+#    >>> Source: https://pytorch-tutorial-assets.s3.amazonaws.com/steam-train-whistle-daniel_simon.mp3
+#    >>> Fetched 8192 bytes.
+#    >>> AudioMetaData(sample_rate=44100, num_frames=0, num_channels=2, bits_per_sample=0, encoding=MP3)

 ######################################################################
 # Loading audio data into Tensor
@@ -411,7 +419,6 @@ inspect_file(path)
 waveform, sample_rate = get_sample(resample=8000)

 formats = [
-    "mp3",
    "flac",
    "vorbis",
    "sph",

--- a/third_party/CMakeLists.txt
+++ b/third_party/CMakeLists.txt
@@ -3,11 +3,8 @@ set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fvisibility=hidden")
 ################################################################################
 # sox
 ################################################################################
-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()

 ################################################################################

--- a/third_party/sox/CMakeLists.txt
+++ b/third_party/sox/CMakeLists.txt
@@ -17,22 +17,24 @@ set(envs
  "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
-)
+if (BUILD_MAD)
+  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
+  )
+endif (BUILD_MAD)

 ExternalProject_Add(amr
  PREFIX ${CMAKE_CURRENT_BINARY_DIR}
@@ -166,7 +168,6 @@ set(SOX_OPTIONS
  --with-amrwb
  --with-flac
  --with-lame
-  --with-mad
  --with-oggvorbis
  --with-opus
  --without-alsa
@@ -190,7 +191,6 @@ 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
@@ -201,9 +201,37 @@ set(SOX_LIBRARIES
  ${INSTALL_DIR}/lib/libogg.a
  )

+set(sox_depends
+  ogg flac vorbis opusfile lame amr
+  )
+
+if (BUILD_MAD)
+  list(
+    APPEND
+    SOX_OPTIONS
+    --with-mad
+    )
+  list(
+    APPEND
+    SOX_LIBRARIES
+    ${INSTALL_DIR}/lib/libmad.a
+    )
+  list(
+    APPEND
+    sox_depends
+    mad
+    )
+else ()
+  list(
+    APPEND
+    SOX_OPTIONS
+    --without-mad
+    )  
+endif (BUILD_MAD)
+
 ExternalProject_Add(sox
  PREFIX ${CMAKE_CURRENT_BINARY_DIR}
-  DEPENDS ogg flac vorbis opusfile lame mad amr
+  DEPENDS ${sox_depends}
  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
@@ -220,6 +248,7 @@ ExternalProject_Add(sox
  LOG_OUTPUT_ON_FAILURE ON
 )

+add_library(libsox INTERFACE)
 add_dependencies(libsox sox)
-set(SOX_INCLUDE_DIR ${INSTALL_DIR}/include PARENT_SCOPE)
-set(SOX_LIBRARIES ${SOX_LIBRARIES} PARENT_SCOPE)
+target_include_directories(libsox INTERFACE ${INSTALL_DIR}/include)
+target_link_libraries(libsox INTERFACE ${SOX_LIBRARIES})
--- a/tools/setup_helpers/extension.py
+++ b/tools/setup_helpers/extension.py
@@ -33,6 +33,7 @@ def _get_build(var, default=False):


 _BUILD_SOX = False if platform.system() == "Windows" else _get_build("BUILD_SOX", True)
+_BUILD_MAD = _get_build("BUILD_MAD", False)
 _BUILD_KALDI = False if platform.system() == "Windows" else _get_build("BUILD_KALDI", True)
 _BUILD_RNNT = _get_build("BUILD_RNNT", True)
 _BUILD_CTC_DECODER = False if platform.system() == "Windows" else _get_build("BUILD_CTC_DECODER", True)
@@ -95,6 +96,7 @@ class CMakeBuild(build_ext):
            "-DCMAKE_VERBOSE_MAKEFILE=ON",
            f"-DPython_INCLUDE_DIR={distutils.sysconfig.get_python_inc()}",
            f"-DBUILD_SOX:BOOL={'ON' if _BUILD_SOX else 'OFF'}",
+            f"-DBUILD_MAD:BOOL={'ON' if _BUILD_MAD else 'OFF'}",
            f"-DBUILD_FFMPEG:BOOL={'ON' if _BUILD_FFMPEG else 'OFF'}",
            f"-DBUILD_KALDI:BOOL={'ON' if _BUILD_KALDI else 'OFF'}",
            f"-DBUILD_RNNT:BOOL={'ON' if _BUILD_RNNT else 'OFF'}",