separate the sox and soxbindings with the requirements

paddlespeech/s2t/frontend/audio.py

@@ -21,7 +21,6 @@ import struct
 import numpy as np
 import resampy
 import soundfile
-import soxbindings as sox
 from scipy import signal
 
 from .utility import convert_samples_from_float32
@@ -355,6 +354,19 @@ class AudioSegment():
         # self._samples = np.interp(new_indices, old_indices, self._samples)
 
         # sox, slow
+        try:
+            import soxbindings as sox
+        except:
+            try:
+                from paddlespeech.s2t.utils import dynamic_pip_install
+                package = "sox"
+                dynamic_pip_install.install(package)
+                package = "soxbindings"
+                dynamic_pip_install.install(package)
+                import soxbindings as sox
+            except:
+                raise RuntimeError("Can not install soxbindings on your system." )
+
         tfm = sox.Transformer()
         tfm.set_globals(multithread=False)
         tfm.speed(speed_rate)

paddlespeech/s2t/transform/perturb.py

@@ -16,7 +16,6 @@ import librosa
 import numpy
 import scipy
 import soundfile
-import soxbindings as sox
 
 from paddlespeech.s2t.io.reader import SoundHDF5File
 
@@ -146,6 +145,19 @@ class SpeedPerturbationSox():
         self.keep_length = keep_length
         self.state = numpy.random.RandomState(seed)
 
+        try:
+            import soxbindings as sox
+        except:
+            try:
+                from paddlespeech.s2t.utils import dynamic_pip_install
+                package = "sox"
+                dynamic_pip_install.install(package)
+                package = "soxbindings"
+                dynamic_pip_install.install(package)
+                import soxbindings as sox
+            except:
+                raise RuntimeError("Can not install soxbindings on your system." )
+
         if utt2ratio is not None:
             self.utt2ratio = {}
             # Use the scheduled ratio for each utterances

setup.py

@@ -55,8 +55,6 @@ requirements = {
         "scipy",
         "sentencepiece~=0.1.96",
         "soundfile~=0.10",
-        "sox",
-        "soxbindings",
         "textgrid",
         "timer",
         "tqdm",
@@ -74,6 +72,8 @@ requirements = {
         "Pillow",
         "pybind11",
         "snakeviz",
+        "sox",
+        "soxbindings",
         "unidecode",
         "yq",
         "pre-commit",