Import torchaudio #1066 4406a6b

Summary: Import up to #1066

Reviewed By: cpuhrsch

Differential Revision: D25373068

fbshipit-source-id: 890d36a25259b93428b3037c3123ff5a2cacfa04

Author: mthrok

.circleci/build_docs/commit_docs.sh

@@ -13,17 +13,19 @@ fi
 src=$1
 target=$2
 
-set -ex
 echo "committing docs from ${src} to ${target}"
 
 git checkout gh-pages
-rm -rf docs/"${target}"/*
-cp -r "${src}"/build/html/* docs/"${target}"
+mkdir -p ./"${target}"
+rm -rf ./"${target}"/*
+cp -r "${src}/build/html/"* ./"$target"
 if [ "${target}" == "master" ]; then
-    rm -rf docs/_static/*
-    cp -r "${src}"/build/html/_static/* docs/_static
+    mkdir -p ./_static
+    rm -rf ./_static/*
+    cp -r "${src}/build/html/_static/"* ./_static
+    git add --all ./_static || true
 fi
-git add docs || true
+git add --all ./"${target}" || true
 git config user.email "[email protected]"
 git config user.name "pytorchbot"
 # If there aren't changes, don't make a commit; push is no-op

.circleci/config.yml

@@ -684,6 +684,16 @@ jobs:
       - attach_workspace:
           at: ~/workspace
       - designate_upload_channel
+      - run:
+          name: Generate netrc
+          command: |
+            # set credentials for https pushing
+            # requires the org-member context
+            cat > ~/.netrc \<<DONE
+              machine github.com
+              login pytorchbot
+              password ${GITHUB_PYTORCHBOT_TOKEN}
+            DONE
       - checkout
       - run:
           name: Upload docs
@@ -785,6 +795,7 @@ workflows:
           requires:
           - binary_linux_wheel_py3.8
       - upload_docs:
+          context: org-member
           filters:
             branches:
               only:

.circleci/config.yml.in

@@ -684,6 +684,16 @@ jobs:
       - attach_workspace:
           at: ~/workspace
       - designate_upload_channel
+      - run:
+          name: Generate netrc
+          command: |
+            # set credentials for https pushing
+            # requires the org-member context
+            cat > ~/.netrc \<<DONE
+              machine github.com
+              login pytorchbot
+              password ${GITHUB_PYTORCHBOT_TOKEN}
+            DONE
       - checkout
       - run:
           name: Upload docs

.circleci/regenerate.py

@@ -79,7 +79,7 @@ def build_doc_job(filter_branch):
     job = {
         "name": "build_docs",
         "python_version": "3.8",
-        "requires": ["binary_linux_wheel_py3.8"],
+        "requires": ["binary_linux_wheel_py3.8", ],
     }
 
     if filter_branch:
@@ -90,8 +90,9 @@ def build_doc_job(filter_branch):
 def upload_doc_job(filter_branch):
     job = {
         "name": "upload_docs",
+        "context": "org-member",
         "python_version": "3.8",
-        "requires": ["build_docs"],
+        "requires": ["build_docs", ],
     }
 
     if filter_branch:

README.md

@@ -10,29 +10,29 @@ the audio domain. By supporting PyTorch, torchaudio follows the same philosophy
 of providing strong GPU acceleration, having a focus on trainable features through
 the autograd system, and having consistent style (tensor names and dimension names).
 Therefore, it is primarily a machine learning library and not a general signal
-processing library. The benefits of PyTorch is be seen in torchaudio through
+processing library. The benefits of PyTorch can be seen in torchaudio through
 having all the computations be through PyTorch operations which makes it easy
 to use and feel like a natural extension.
 
-- [Support audio I/O (Load files, Save files)](http://pytorch.org/audio/)
+- [Support audio I/O (Load files, Save files)](http://pytorch.org/audio/stable/)
   - Load the following formats into a torch Tensor using SoX
     - mp3, wav, aac, ogg, flac, avr, cdda, cvs/vms,
     - aiff, au, amr, mp2, mp4, ac3, avi, wmv,
     - mpeg, ircam and any other format supported by libsox.
-    - [Kaldi (ark/scp)](http://pytorch.org/audio/kaldi_io.html)
-- [Dataloaders for common audio datasets (VCTK, YesNo)](http://pytorch.org/audio/datasets.html)
+    - [Kaldi (ark/scp)](http://pytorch.org/audio/stable/kaldi_io.html)
+- [Dataloaders for common audio datasets (VCTK, YesNo)](http://pytorch.org/audio/stable/datasets.html)
 - Common audio transforms
-    - [Spectrogram, AmplitudeToDB, MelScale, MelSpectrogram, MFCC, MuLawEncoding, MuLawDecoding, Resample](http://pytorch.org/audio/transforms.html)
+    - [Spectrogram, AmplitudeToDB, MelScale, MelSpectrogram, MFCC, MuLawEncoding, MuLawDecoding, Resample](http://pytorch.org/audio/stable/transforms.html)
 - Compliance interfaces: Run code using PyTorch that align with other libraries
-    - [Kaldi: spectrogram, fbank, mfcc, resample_waveform](https://pytorch.org/audio/compliance.kaldi.html)
+    - [Kaldi: spectrogram, fbank, mfcc, resample_waveform](https://pytorch.org/audio/stable/compliance.kaldi.html)
 
 Dependencies
 ------------
 * PyTorch (See below for the compatible versions)
 * libsox v14.3.2 or above (only required when building from source)
 * [optional] vesis84/kaldi-io-for-python commit cb46cb1f44318a5d04d4941cf39084c5b021241e or above
 
-The following is the corresponding ``torchaudio`` versions and supported Python versions.
+The following are the corresponding ``torchaudio`` versions and supported Python versions.
 
 | ``torch``                | ``torchaudio``           | ``python``                      |
 | ------------------------ | ------------------------ | ------------------------------- |
@@ -46,7 +46,7 @@ The following is the corresponding ``torchaudio`` versions and supported Python
 Installation
 ------------
 
-### Binary Distibutions
+### Binary Distributions
 
 To install the latest version using anaconda, run:
 
@@ -127,7 +127,7 @@ BUILD_SOX=1 MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py i
 ```
 
 This is known to work on linux and unix distributions such as Ubuntu and CentOS 7 and macOS.
-If you try this on a new system and found a solution to make it work, feel free to share it by opening and issue.
+If you try this on a new system and find a solution to make it work, feel free to share it by opening an issue.
 
 #### Troubleshooting
 
@@ -195,16 +195,16 @@ Conventions
 
 With torchaudio being a machine learning library and built on top of PyTorch,
 torchaudio is standardized around the following naming conventions. Tensors are
-assumed to have channel as the first dimension and time as the last
+assumed to have channels as the first dimension and time as the last
 dimension (when applicable). This makes it consistent with PyTorch's dimensions.
 For size names, the prefix `n_` is used (e.g. "a tensor of size (`n_freq`, `n_mel`)")
 whereas dimension names do not have this prefix (e.g. "a tensor of
-dimension (channel, time)")
+dimension (channels, time)")
 
-* `waveform`: a tensor of audio samples with dimensions (channel, time)
+* `waveform`: a tensor of audio samples with dimensions (channels, time)
 * `sample_rate`: the rate of audio dimensions (samples per second)
-* `specgram`: a tensor of spectrogram with dimensions (channel, freq, time)
-* `mel_specgram`: a mel spectrogram with dimensions (channel, mel, time)
+* `specgram`: a tensor of spectrogram with dimensions (channels, freq, time)
+* `mel_specgram`: a mel spectrogram with dimensions (channels, mel, time)
 * `hop_length`: the number of samples between the starts of consecutive frames
 * `n_fft`: the number of Fourier bins
 * `n_mel`, `n_mfcc`: the number of mel and MFCC bins
@@ -216,16 +216,16 @@ dimension (channel, time)")
 
 Transforms expect and return the following dimensions.
 
-* `Spectrogram`: (channel, time) -> (channel, freq, time)
-* `AmplitudeToDB`: (channel, freq, time) -> (channel, freq, time)
-* `MelScale`: (channel, freq, time) -> (channel, mel, time)
-* `MelSpectrogram`: (channel, time) -> (channel, mel, time)
-* `MFCC`: (channel, time) -> (channel, mfcc, time)
-* `MuLawEncode`: (channel, time) -> (channel, time)
-* `MuLawDecode`: (channel, time) -> (channel, time)
-* `Resample`: (channel, time) -> (channel, time)
-* `Fade`: (channel, time) -> (channel, time)
-* `Vol`: (channel, time) -> (channel, time)
+* `Spectrogram`: (channels, time) -> (channels, freq, time)
+* `AmplitudeToDB`: (channels, freq, time) -> (channels, freq, time)
+* `MelScale`: (channels, freq, time) -> (channels, mel, time)
+* `MelSpectrogram`: (channels, time) -> (channels, mel, time)
+* `MFCC`: (channels, time) -> (channel, mfcc, time)
+* `MuLawEncode`: (channels, time) -> (channels, time)
+* `MuLawDecode`: (channels, time) -> (channels, time)
+* `Resample`: (channels, time) -> (channels, time)
+* `Fade`: (channels, time) -> (channels, time)
+* `Vol`: (channels, time) -> (channels, time)
 
 Complex numbers are supported via tensors of dimension (..., 2), and torchaudio provides `complex_norm` and `angle` to convert such a tensor into its magnitude and phase. Here, and in the documentation, we use an ellipsis "..." as a placeholder for the rest of the dimensions of a tensor, e.g. optional batching and channel dimensions.
 

build_tools/setup_helpers/extension.py

@@ -89,6 +89,8 @@ def _get_extra_objects():
             'libvorbisfile.a',
             'libvorbis.a',
             'libogg.a',
+            'libopencore-amrnb.a',
+            'libopencore-amrwb.a',
         ]
         for lib in libs:
             objs.append(str(_TP_INSTALL_DIR / 'lib' / lib))

docs/source/datasets.rst

@@ -2,7 +2,7 @@ torchaudio.datasets
 ====================
 
 All datasets are subclasses of :class:`torch.utils.data.Dataset`
-i.e, they have ``__getitem__`` and ``__len__`` methods implemented.
+and have ``__getitem__`` and ``__len__`` methods implemented.
 Hence, they can all be passed to a :class:`torch.utils.data.DataLoader`
 which can load multiple samples parallelly using ``torch.multiprocessing`` workers.
 For example: ::

docs/source/functional.rst

@@ -124,11 +124,21 @@ vad
 
 .. autofunction:: spectrogram
 
+:hidden:`griffinlim`
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: griffinlim
+
 :hidden:`amplitude_to_DB`
 ~~~~~~~~~~~~~~~~~~~~~~~~~
 
 .. autofunction:: amplitude_to_DB
 
+:hidden:`DB_to_amplitude`
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autofunction:: DB_to_amplitude
+
 :hidden:`create_fb_matrix`
 ~~~~~~~~~~~~~~~~~~~~~~~~~~
 

test/torchaudio_unittest/datasets/librispeech_test.py

@@ -1,4 +1,5 @@
 import os
+from pathlib import Path
 
 from torchaudio.datasets import librispeech
 
@@ -91,11 +92,7 @@ def tearDownClass(cls):
         # In case of test failure
         librispeech.LIBRISPEECH._ext_audio = '.flac'
 
-    def test_librispeech(self):
-        librispeech.LIBRISPEECH._ext_audio = '.wav'
-        dataset = librispeech.LIBRISPEECH(self.root_dir)
-        print(dataset._path)
-
+    def _test_librispeech(self, dataset):
         num_samples = 0
         for i, (
             data, sample_rate, utterance, speaker_id, chapter_id, utterance_id
@@ -110,3 +107,13 @@ def test_librispeech(self):
 
         assert num_samples == len(self.samples)
         librispeech.LIBRISPEECH._ext_audio = '.flac'
+
+    def test_librispeech_str(self):
+        librispeech.LIBRISPEECH._ext_audio = '.wav'
+        dataset = librispeech.LIBRISPEECH(self.root_dir)
+        self._test_librispeech(dataset)
+
+    def test_librispeech_path(self):
+        librispeech.LIBRISPEECH._ext_audio = '.wav'
+        dataset = librispeech.LIBRISPEECH(Path(self.root_dir))
+        self._test_librispeech(dataset)

test/torchaudio_unittest/datasets/libritts_test.py

@@ -1,4 +1,5 @@
 import os
+from pathlib import Path
 
 from torchaudio.datasets.libritts import LIBRITTS
 
@@ -47,8 +48,7 @@ def setUpClass(cls):
             with open(path_normalized, 'w') as file_:
                 file_.write(cls.normalized_text)
 
-    def test_libritts(self):
-        dataset = LIBRITTS(self.root_dir)
+    def _test_libritts(self, dataset):
         n_ites = 0
         for i, (waveform,
                 sample_rate,
@@ -69,3 +69,11 @@ def test_libritts(self):
             assert utterance_id == f'{"_".join(str(u) for u in expected_ids[-4:])}'
             n_ites += 1
         assert n_ites == len(self.utterance_ids)
+
+    def test_libritts_str(self):
+        dataset = LIBRITTS(self.root_dir)
+        self._test_libritts(dataset)
+
+    def test_libritts_path(self):
+        dataset = LIBRITTS(Path(self.root_dir))
+        self._test_libritts(dataset)