include format in frontend io

frontend/io.py

@@ -468,7 +468,9 @@ def read_audio(input_file_name, framerate=None):
 def write_label(label,
                 output_file_name,
                 selected_label='speech',
-                frame_per_second=100):
+                frame_per_second=100,
+                show=None,
+                format="mdtm"):
     """Save labels in ALIZE format
 
     :param output_file_name: name of the file to write to
@@ -483,11 +485,29 @@ def write_label(label,
         # append 0 at the beginning of the list, append the last index to the list
         idx = [0] + (numpy.arange(len(bits))[bits] + 1).tolist() + [len(label)]
         framerate = decimal.Decimal(1) / decimal.Decimal(frame_per_second)
-        # for each pair of indexes (idx[i] and idx[i+1]), create a segment
-        with open(output_file_name, 'w') as fid:
+
+        if format == "lab":
+            # for each pair of indexes (idx[i] and idx[i+1]), create a segment
+            with open(output_file_name, 'w') as fid:
+                for i in range(~label[0], len(idx) - 1, 2):
+                    fid.write('{} {} {}\n'.format(str(idx[i]*framerate),
+                                                  str(idx[i + 1]*framerate), selected_label))
+        else:
+            # write in MDTM format
+            lst = []
             for i in range(~label[0], len(idx) - 1, 2):
-                fid.write('{} {} {}\n'.format(str(idx[i]*framerate),
-                                              str(idx[i + 1]*framerate), selected_label))
+                gender = 'U'
+                env = 'U'
+                channel = 'U'
+                start = idx[i]*framerate
+                stop = idx[i + 1]*framerate
+                lst.append('{:s} 1 {:.2f} {:.2f} {:s} {:s} {:s} {:s}\n'.format(
+                    show, start, stop - start, gender,
+                    channel, env, "speech"))
+
+            with open(output_file_name, 'w', encoding="utf8") as fid:
+                for line in lst:
+                    fid.write(line)
 
 
 def read_label(input_file_name, selected_label='speech', frame_per_second=100):

nnet/sincnet.py

@@ -337,7 +337,6 @@ class SincNet(torch.nn.Module):
                     bias=True,
                 )
             else:
-                print(f"smaple_rate ={self.sample_rate} ")
                 conv1d = SincConv1d(
                                     out_channels,
                                     kernel_size,