Update Tacotron2 docs (#1840)

docs/source/models.rst

@@ -31,8 +31,8 @@ Tacotron2
 Model
 -----
 
-Tacotoron2
-^^^^^^^^^^
+Tacotron2
+^^^^^^^^^
 
 .. autoclass:: Tacotron2
 

torchaudio/models/tacotron2.py

@@ -1091,21 +1091,23 @@ class Tacotron2(nn.Module):
         The input ``mel_specgram`` should be padded with zeros to length max of ``mel_specgram_lengths``.
 
         Args:
-            text (Tensor): The input text to Tacotron2 with shape (n_batch, max of ``text_lengths``).
-            text_lengths (Tensor): The length of each text with shape (n_batch).
+            text (Tensor): The input text to Tacotron2 with shape `(n_batch, max of text_lengths)`.
+            text_lengths (Tensor): The length of each text with shape `(n_batch, )`.
             mel_specgram (Tensor): The target mel spectrogram
-                with shape (n_batch, n_mels, max of ``mel_specgram_lengths``).
-            mel_specgram_lengths (Tensor): The length of each mel spectrogram with shape (n_batch).
+                with shape `(n_batch, n_mels, max of mel_specgram_lengths)`.
+            mel_specgram_lengths (Tensor): The length of each mel spectrogram with shape `(n_batch, )`.
 
         Returns:
-            mel_specgram (Tensor): Mel spectrogram before Postnet
-                with shape (n_batch, n_mels, max of ``mel_specgram_lengths``).
-            mel_specgram_postnet (Tensor): Mel spectrogram after Postnet
-                with shape (n_batch, n_mels, max of ``mel_specgram_lengths``).
-            stop_token (Tensor): The output for stop token at each time step
-                with shape (n_batch, max of ``mel_specgram_lengths``).
-            alignment (Tensor): Sequence of attention weights from the decoder.
-                with shape (n_batch, max of ``mel_specgram_lengths``, max of ``text_lengths``).
+            Tensor, Tensor, Tensor, and Tensor:
+                Tensor
+                    Mel spectrogram before Postnet with shape `(n_batch, n_mels, max of mel_specgram_lengths)`.
+                Tensor
+                    Mel spectrogram after Postnet with shape `(n_batch, n_mels, max of mel_specgram_lengths)`.
+                Tensor
+                    The output for stop token at each time step with shape `(n_batch, max of mel_specgram_lengths)`.
+                Tensor
+                    Sequence of attention weights from the decoder with
+                    shape `(n_batch, max of mel_specgram_lengths, max of text_lengths)`.
         """
 
         embedded_inputs = self.embedding(text).transpose(1, 2)
@@ -1139,17 +1141,19 @@ class Tacotron2(nn.Module):
         The input `text` should be padded with zeros to length max of ``text_lengths``.
 
         Args:
-            text (Tensor): The input text to Tacotron2 with shape (n_batch, max of ``text_lengths``).
+            text (Tensor): The input text to Tacotron2 with shape `(n_batch, max of text_lengths)`.
             text_lengths (Tensor or None, optional): The length of each text with shape `(n_batch, )`.
                 If ``None``, it is assumed that the all the texts are valid. Default: ``None``
 
-        Return:
-            mel_specgram (Tensor): The predicted mel spectrogram
-                with shape (n_batch, n_mels, max of ``mel_specgram_lengths.max()``).
-            mel_specgram_lengths (Tensor): The length of the predicted mel spectrogram
-                with shape (n_batch, ).
-            alignments (Tensor): Sequence of attention weights from the decoder.
-                with shape (n_batch, max of ``mel_specgram_lengths``, max of ``text_lengths``).
+        Returns:
+            Tensor, Tensor, and Tensor:
+                Tensor
+                    The predicted mel spectrogram with shape `(n_batch, n_mels, max of mel_specgram_lengths)`.
+                Tensor
+                    The length of the predicted mel spectrogram with shape `(n_batch, )`.
+                Tensor
+                    Sequence of attention weights from the decoder with shape
+                    `(n_batch, max of mel_specgram_lengths, max of text_lengths)`.
         """
         n_batch, max_length = text.shape
         if text_lengths is None: