Add sox_io_backend

test/sox_io/test_torchscript.py

@@ -2,7 +2,7 @@
 from typing import Optional
 
 import torch
-from torchaudio.backend import sox_io_backend
+import torchaudio
 from parameterized import parameterized
 
 from .. import common_utils
@@ -19,11 +19,11 @@
 
 
 def py_info_func(filepath: str) -> torch.classes.torchaudio.SignalInfo:
-    return sox_io_backend.info(filepath)
+    return torchaudio.info(filepath)
 
 
 def py_load_func(filepath: str):
-    return sox_io_backend.load(filepath)
+    return torchaudio.load(filepath)
 
 
 def py_save_func(
@@ -33,12 +33,14 @@ def py_save_func(
         channel_first: bool = False,
         compression: Optional[float] = None,
 ):
-    sox_io_backend.save(filepath, tensor, sample_rate, channel_first, compression)
+    torchaudio.save(filepath, tensor, sample_rate, channel_first, compression)
 
 
 @common_utils.skipIfNoExec('sox')
 @common_utils.skipIfNoExtension
 class SoxIO(TempDirMixin, TorchaudioTestCase):
+    backend = 'sox_io'
+
     @parameterized.expand(list(itertools.product(
         ['float32', 'int32', 'int16', 'uint8'],
         [8000, 16000],

test/test_backend.py

@@ -1,5 +1,3 @@
-import unittest
-
 import torchaudio
 
 from . import common_utils
@@ -33,6 +31,12 @@ class TestBackendSwitch_SoX(BackendSwitchMixin, common_utils.TorchaudioTestCase)
     backend_module = torchaudio.backend.sox_backend
 
 
+@common_utils.skipIfNoExtension
+class TestBackendSwitch_SoXIO(BackendSwitchMixin, common_utils.TorchaudioTestCase):
+    backend = 'sox_io'
+    backend_module = torchaudio.backend.sox_io_backend
+
+
 @common_utils.skipIfNoModule('soundfile')
 class TestBackendSwitch_soundfile(BackendSwitchMixin, common_utils.TorchaudioTestCase):
     backend = 'soundfile'

test/test_io.py

@@ -17,11 +17,15 @@ class Test_LoadSave(unittest.TestCase):
 
     def test_1_save(self):
         for backend in BACKENDS_MP3:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_1_save(self.test_filepath, False)
 
         for backend in BACKENDS:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_1_save(self.test_filepath_wav, True)
@@ -68,6 +72,8 @@ def _test_1_save(self, test_filepath, normalization):
 
     def test_1_save_sine(self):
         for backend in BACKENDS:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_1_save_sine()
@@ -101,11 +107,15 @@ def _test_1_save_sine(self):
 
     def test_2_load(self):
         for backend in BACKENDS_MP3:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_2_load(self.test_filepath, 278756)
 
         for backend in BACKENDS:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_2_load(self.test_filepath_wav, 276858)
@@ -142,6 +152,8 @@ def _test_2_load(self, test_filepath, length):
 
     def test_2_load_nonormalization(self):
         for backend in BACKENDS_MP3:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_2_load_nonormalization(self.test_filepath, 278756)
@@ -159,6 +171,8 @@ def _test_2_load_nonormalization(self, test_filepath, length):
 
     def test_3_load_and_save_is_identity(self):
         for backend in BACKENDS:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_3_load_and_save_is_identity()
@@ -197,6 +211,8 @@ def _test_3_load_and_save_is_identity_across_backend(self, backend1, backend2):
 
     def test_4_load_partial(self):
         for backend in BACKENDS_MP3:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_4_load_partial()
@@ -239,6 +255,8 @@ def _test_4_load_partial(self):
 
     def test_5_get_info(self):
         for backend in BACKENDS:
+            if backend == 'sox_io':
+                continue
             with self.subTest():
                 torchaudio.set_audio_backend(backend)
                 self._test_5_get_info()

torchaudio/backend/utils.py

@@ -7,6 +7,7 @@
 from . import (
     no_backend,
     sox_backend,
+    sox_io_backend,
     soundfile_backend,
 )
 
@@ -24,6 +25,7 @@ def list_audio_backends() -> List[str]:
         backends.append('soundfile')
     if is_module_available('torchaudio._torchaudio'):
         backends.append('sox')
+        backends.append('sox_io')
     return backends
 
 
@@ -43,6 +45,8 @@ def set_audio_backend(backend: Optional[str]) -> None:
         module = no_backend
     elif backend == 'sox':
         module = sox_backend
+    elif backend == 'sox_io':
+        module = sox_io_backend
     elif backend == 'soundfile':
         module = soundfile_backend
     else:
@@ -69,6 +73,8 @@ def get_audio_backend() -> Optional[str]:
         return None
     if torchaudio.load == sox_backend.load:
         return 'sox'
+    if torchaudio.load == sox_io_backend.load:
+        return 'sox_io'
     if torchaudio.load == soundfile_backend.load:
         return 'soundfile'
     raise ValueError('Unknown backend.')

torchaudio/csrc/sox_io.cpp

@@ -31,6 +31,46 @@ struct SoxDescriptor {
   sox_format_t* fd_;
 };
 
+void printHighPrecision(const torch::Tensor& t) {
+  AT_DISPATCH_ALL_TYPES(t.scalar_type(), "print", [&] {
+    auto acc = t.accessor<scalar_t, 2>();
+    for (int i = 0; i < acc.size(0); i++) {
+      for (int j = 0; j < acc.size(1); j++) {
+        auto val = acc[i][j];
+        std::cout << val << " ";
+      }
+      std::cout << std::endl;
+    }
+  });
+}
+
+void printEnds(const torch::Tensor& t) {
+  std::cout << "++ first 10" << std::endl;
+  printHighPrecision(t.index({Slice(None, 10, None)}));
+  std::cout << "++ last 10" << std::endl;
+  printHighPrecision(t.index({Slice(-10, None, None)}));
+}
+
+void printSignalInfo(const sox_signalinfo_t signal) {
+  std::cout << " - rate: " << signal.rate << std::endl
+            << " - channels: " << signal.channels << std::endl
+            << " - precision: " << signal.precision << std::endl
+            << " - length: " << signal.length << std::endl;
+  if (signal.mult) {
+    std::cout << " - mult: " << *(signal.mult) << std::endl;
+  }
+}
+
+void printEncodingInfo(const sox_encodinginfo_t encoding) {
+  std::cout << " - encoding: " << encoding.encoding << std::endl
+            << " - bits_per_sample: " << encoding.bits_per_sample << std::endl
+            << " - compression: " << encoding.compression << std::endl
+            << " - reverse_bytes: " << encoding.reverse_bytes << std::endl
+            << " - reverse_nibbles: " << encoding.reverse_nibbles << std::endl
+            << " - reverse_bits: " << encoding.reverse_bits << std::endl
+            << " - opposite_endian: " << encoding.opposite_endian << std::endl;
+}
+
 } // namespace
 
 c10::intrusive_ptr<::torchaudio::SignalInfo> get_info(
@@ -45,6 +85,9 @@ c10::intrusive_ptr<::torchaudio::SignalInfo> get_info(
     throw std::runtime_error("Error opening audio file");
   }
 
+  printSignalInfo(sd->signal);
+  printEncodingInfo(sd->encoding);
+
   return c10::make_intrusive<::torchaudio::SignalInfo>(
       static_cast<int64_t>(sd->signal.rate),
       static_cast<int64_t>(sd->signal.channels),
@@ -186,6 +229,7 @@ torch::Tensor load_audio_file(
       option = option.dtype(torch::kFloat32);
       break;
     default:
+      std::cout << "encoding: " << ei.encoding << std::endl;
       throw std::runtime_error("Unsupported encoding.");
   }
   tensor = tensor.to(option);
@@ -313,6 +357,10 @@ void save_audio_file(
                                          /*reverse_bits=*/sox_option_default,
                                          /*opposite_endian=*/sox_false};
 
+  // std::cout << "++ Input info: " << std::endl;
+  // printSignalInfo(signal_info);
+  // printEncodingInfo(encoding_info);
+
   SoxDescriptor sd(sox_open_write(
       file_name.c_str(),
       &signal_info,
@@ -325,7 +373,13 @@ void save_audio_file(
     throw std::runtime_error("Error saving audio file: failed to open file.");
   }
 
+  std::cout << "++ Detected info: " << std::endl;
+  printSignalInfo(sd->signal);
+  printEncodingInfo(sd->encoding);
+
   auto tensor_ = tensor;
+  // std::cout << "++ tensor before normalization" << std::endl;
+  // printEnds(tensor_);
 
   // de-normalization
   if (dtype == torch::kFloat32) {
@@ -343,6 +397,8 @@ void save_audio_file(
     tensor_ -= 128;
     tensor_ *= 16777216;
   }
+  // std::cout << "++ tensor after nomalization" << std::endl;
+  // printEnds(tensor_);
 
   // Format & clean up
   if (channel_first) {