Merge pull request #13 from fumiama/mono

feat: drop cython monotonic_align

.gitignore

@@ -158,3 +158,5 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
+
+/models

bert/chinese-roberta-wwm-ext-large/.gitignore

@@ -0,0 +1 @@
+*.bin

monotonic_align/__init__.py

@@ -1,19 +1,15 @@
-import numpy as np
+from numpy import zeros, int32, float32
-import torch
+from torch import from_numpy
-from .monotonic_align.core import maximum_path_c
 
+from .core import maximum_path_jit
 
 def maximum_path(neg_cent, mask):
-  """ Cython optimized version.
-  neg_cent: [b, t_t, t_s]
-  mask: [b, t_t, t_s]
-  """
   device = neg_cent.device
   dtype = neg_cent.dtype
-  neg_cent = neg_cent.data.cpu().numpy().astype(np.float32)
+  neg_cent = neg_cent.data.cpu().numpy().astype(float32)
-  path = np.zeros(neg_cent.shape, dtype=np.int32)
+  path = zeros(neg_cent.shape, dtype=int32)
 
-  t_t_max = mask.sum(1)[:, 0].data.cpu().numpy().astype(np.int32)
+  t_t_max = mask.sum(1)[:, 0].data.cpu().numpy().astype(int32)
-  t_s_max = mask.sum(2)[:, 0].data.cpu().numpy().astype(np.int32)
+  t_s_max = mask.sum(2)[:, 0].data.cpu().numpy().astype(int32)
-  maximum_path_c(path, neg_cent, t_t_max, t_s_max)
+  maximum_path_jit(path, neg_cent, t_t_max, t_s_max)
-  return torch.from_numpy(path).to(device=device, dtype=dtype)
+  return from_numpy(path).to(device=device, dtype=dtype)

monotonic_align/core.py

@@ -0,0 +1,35 @@
+import numba
+
+
+@numba.jit(numba.void(numba.int32[:,:,::1], numba.float32[:,:,::1], numba.int32[::1], numba.int32[::1]), nopython=True, nogil=True)
+def maximum_path_jit(paths, values, t_ys, t_xs):
+  b = paths.shape[0]
+  max_neg_val=-1e9
+  for i in range(int(b)):
+    path = paths[i]
+    value = values[i]
+    t_y = t_ys[i]
+    t_x = t_xs[i]
+
+    v_prev = v_cur = 0.0
+    index = t_x - 1
+
+    for y in range(t_y):
+      for x in range(max(0, t_x + y - t_y), min(t_x, y + 1)):
+        if x == y:
+          v_cur = max_neg_val
+        else:
+          v_cur = value[y-1, x]
+        if x == 0:
+          if y == 0:
+            v_prev = 0.
+          else:
+            v_prev = max_neg_val
+        else:
+          v_prev = value[y-1, x-1]
+        value[y, x] += max(v_prev, v_cur)
+
+    for y in range(t_y - 1, -1, -1):
+      path[y, index] = 1
+      if index != 0 and (index == y or value[y-1, index] < value[y-1, index-1]):
+        index = index - 1

monotonic_align/core.pyx

@@ -1,42 +0,0 @@
-cimport cython
-from cython.parallel import prange
-
-
-@cython.boundscheck(False)
-@cython.wraparound(False)
-cdef void maximum_path_each(int[:,::1] path, float[:,::1] value, int t_y, int t_x, float max_neg_val=-1e9) nogil:
-  cdef int x
-  cdef int y
-  cdef float v_prev
-  cdef float v_cur
-  cdef float tmp
-  cdef int index = t_x - 1
-
-  for y in range(t_y):
-    for x in range(max(0, t_x + y - t_y), min(t_x, y + 1)):
-      if x == y:
-        v_cur = max_neg_val
-      else:
-        v_cur = value[y-1, x]
-      if x == 0:
-        if y == 0:
-          v_prev = 0.
-        else:
-          v_prev = max_neg_val
-      else:
-        v_prev = value[y-1, x-1]
-      value[y, x] += max(v_prev, v_cur)
-
-  for y in range(t_y - 1, -1, -1):
-    path[y, index] = 1
-    if index != 0 and (index == y or value[y-1, index] < value[y-1, index-1]):
-      index = index - 1
-
-
-@cython.boundscheck(False)
-@cython.wraparound(False)
-cpdef void maximum_path_c(int[:,:,::1] paths, float[:,:,::1] values, int[::1] t_ys, int[::1] t_xs) nogil:
-  cdef int b = paths.shape[0]
-  cdef int i
-  for i in prange(b, nogil=True):
-    maximum_path_each(paths[i], values[i], t_ys[i], t_xs[i])

monotonic_align/setup.py

@@ -1,9 +0,0 @@
-from distutils.core import setup
-from Cython.Build import cythonize
-import numpy
-
-setup(
-  name = 'monotonic_align',
-  ext_modules = cythonize("core.pyx"),
-  include_dirs=[numpy.get_include()]
-)

requirements.txt

@@ -2,6 +2,7 @@ Cython
 librosa==0.9.1
 matplotlib
 numpy
+numba
 phonemizer
 scipy
 tensorboard