fix some typos

examples/other/1xt2x/src_deepspeech2x/models/ds2/deepspeech2.py

@@ -149,13 +149,13 @@ class DeepSpeech2Model(nn.Layer):
         """Compute Model loss
 
         Args:
-            audio (Tenosr): [B, T, D]
+            audio (Tensor): [B, T, D]
             audio_len (Tensor): [B]
             text (Tensor): [B, U]
             text_len (Tensor): [B]
 
         Returns:
-            loss (Tenosr): [1]
+            loss (Tensor): [1]
         """
         eouts, eouts_len = self.encoder(audio, audio_len)
         loss = self.decoder(eouts, eouts_len, text, text_len)

paddlespeech/s2t/decoders/ctcdecoder/scorer_deprecated.py

@@ -62,7 +62,7 @@ class Scorer(object):
         """Evaluation function, gathering all the different scores
         and return the final one.
 
-        :param sentence: The input sentence for evalutation
+        :param sentence: The input sentence for evaluation
         :type sentence: str
         :param log: Whether return the score in log representation.
         :type log: bool

paddlespeech/s2t/decoders/ctcdecoder/swig/ctc_beam_search_decoder.cpp

@@ -183,7 +183,7 @@ std::vector<std::pair<double, std::string>> ctc_beam_search_decoder(
     std::sort(
         prefixes.begin(), prefixes.begin() + num_prefixes, prefix_compare);
 
-    // compute aproximate ctc score as the return score, without affecting the
+    // compute approximate ctc score as the return score, without affecting the
     // return order of decoding result. To delete when decoder gets stable.
     for (size_t i = 0; i < beam_size && i < prefixes.size(); ++i) {
         double approx_ctc = prefixes[i]->score;

paddlespeech/s2t/decoders/ctcdecoder/swig/decoder_utils.cpp

@@ -26,7 +26,7 @@ std::vector<std::pair<size_t, float>> get_pruned_log_probs(
     for (size_t i = 0; i < prob_step.size(); ++i) {
         prob_idx.push_back(std::pair<int, double>(i, prob_step[i]));
     }
-    // pruning of vacobulary
+    // pruning of vocabulary
     size_t cutoff_len = prob_step.size();
     if (cutoff_prob < 1.0 || cutoff_top_n < cutoff_len) {
         std::sort(prob_idx.begin(),

paddlespeech/s2t/decoders/ctcdecoder/swig/scorer.cpp

@@ -223,7 +223,7 @@ void Scorer::fill_dictionary(bool add_space) {
 
      * This gets rid of "epsilon" transitions in the FST.
      * These are transitions that don't require a string input to be taken.
-     * Getting rid of them is necessary to make the FST determinisitc, but
+     * Getting rid of them is necessary to make the FST deterministic, but
      * can greatly increase the size of the FST
      */
     fst::RmEpsilon(&dictionary);

paddlespeech/s2t/decoders/scorers/ctc.py

@@ -154,7 +154,7 @@ class CTCPrefixScorer(BatchPartialScorerInterface):
         Args:
             state: The states of hyps
 
-        Returns: exteded state
+        Returns: extended state
 
         """
         new_state = []

paddlespeech/s2t/decoders/scorers/ctc_prefix_score.py

@@ -11,7 +11,7 @@ class CTCPrefixScorePD():
 
     which is based on Algorithm 2 in WATANABE et al.
     "HYBRID CTC/ATTENTION ARCHITECTURE FOR END-TO-END SPEECH RECOGNITION,"
-    but extended to efficiently compute the label probablities for multiple
+    but extended to efficiently compute the label probabilities for multiple
     hypotheses simultaneously
     See also Seki et al. "Vectorized Beam Search for CTC-Attention-Based
     Speech Recognition," In INTERSPEECH (pp. 3825-3829), 2019.
@@ -272,7 +272,7 @@ class CTCPrefixScore():
 
     which is based on Algorithm 2 in WATANABE et al.
     "HYBRID CTC/ATTENTION ARCHITECTURE FOR END-TO-END SPEECH RECOGNITION,"
-    but extended to efficiently compute the probablities of multiple labels
+    but extended to efficiently compute the probabilities of multiple labels
     simultaneously
     """
 

paddlespeech/s2t/models/ds2/deepspeech2.py

@@ -151,13 +151,13 @@ class DeepSpeech2Model(nn.Layer):
         """Compute Model loss
 
         Args:
-            audio (Tenosr): [B, T, D]
+            audio (Tensors): [B, T, D]
             audio_len (Tensor): [B]
             text (Tensor): [B, U]
             text_len (Tensor): [B]
 
         Returns:
-            loss (Tenosr): [1]
+            loss (Tensor): [1]
         """
         eouts, eouts_len = self.encoder(audio, audio_len)
         loss = self.decoder(eouts, eouts_len, text, text_len)

paddlespeech/s2t/models/ds2_online/deepspeech2.py

@@ -279,13 +279,13 @@ class DeepSpeech2ModelOnline(nn.Layer):
         """Compute Model loss
 
         Args:
-            audio (Tenosr): [B, T, D]
+            audio (Tensor): [B, T, D]
             audio_len (Tensor): [B]
             text (Tensor): [B, U]
             text_len (Tensor): [B]
 
         Returns:
-            loss (Tenosr): [1]
+            loss (Tensor): [1]
         """
         eouts, eouts_len, final_state_h_box, final_state_c_box = self.encoder(
             audio, audio_len, None, None)

paddlespeech/s2t/models/u2/u2.py

@@ -680,8 +680,8 @@ class U2BaseModel(ASRInterface, nn.Layer):
         """u2 decoding.
 
         Args:
-            feats (Tenosr): audio features, (B, T, D)
+            feats (Tensor): audio features, (B, T, D)
-            feats_lengths (Tenosr): (B)
+            feats_lengths (Tensor): (B)
             text_feature (TextFeaturizer): text feature object.
             decoding_method (str): decoding mode, e.g.
                     'attention', 'ctc_greedy_search',

paddlespeech/s2t/models/u2_st/u2_st.py

@@ -478,8 +478,8 @@ class U2STBaseModel(nn.Layer):
         """u2 decoding.
 
         Args:
-            feats (Tenosr): audio features, (B, T, D)
+            feats (Tensor): audio features, (B, T, D)
-            feats_lengths (Tenosr): (B)
+            feats_lengths (Tensor): (B)
             text_feature (TextFeaturizer): text feature object.
             decoding_method (str): decoding mode, e.g.
                     'fullsentence',

paddlespeech/s2t/modules/ctc.py

@@ -81,10 +81,10 @@ class CTCDecoderBase(nn.Layer):
         Args:
             hs_pad (Tensor): batch of padded hidden state sequences (B, Tmax, D)
             hlens (Tensor): batch of lengths of hidden state sequences (B)
-            ys_pad (Tenosr): batch of padded character id sequence tensor (B, Lmax)
+            ys_pad (Tensor): batch of padded character id sequence tensor (B, Lmax)
             ys_lens (Tensor): batch of lengths of character sequence (B)
         Returns:
-            loss (Tenosr): ctc loss value, scalar.
+            loss (Tensor): ctc loss value, scalar.
         """
         logits = self.ctc_lo(self.dropout(hs_pad))
         loss = self.criterion(logits, ys_pad, hlens, ys_lens)
@@ -252,8 +252,8 @@ class CTCDecoder(CTCDecoderBase):
         """ctc decoding with probs.
 
         Args:
-            probs (Tenosr): activation after softmax
+            probs (Tensor): activation after softmax
-            logits_lens (Tenosr): audio output lens
+            logits_lens (Tensor): audio output lens
             vocab_list ([type]): [description]
             decoding_method ([type]): [description]
             lang_model_path ([type]): [description]

paddlespeech/s2t/modules/mask.py

@@ -54,7 +54,7 @@ def make_pad_mask(lengths: paddle.Tensor) -> paddle.Tensor:
                  [0, 0, 0, 1, 1],
                  [0, 0, 1, 1, 1]]
     """
-    # (TODO: Hui Zhang): jit not support Tenosr.dim() and Tensor.ndim
+    # (TODO: Hui Zhang): jit not support Tensor.dim() and Tensor.ndim
     # assert lengths.dim() == 1
     batch_size = int(lengths.shape[0])
     max_len = int(lengths.max())

paddlespeech/s2t/utils/dynamic_import.py

@@ -57,7 +57,7 @@ def filter_valid_args(args: Dict[Text, Any], valid_keys: List[Text]):
     return new_args
 
 
-def filter_out_tenosr(args: Dict[Text, Any]):
+def filter_out_tensor(args: Dict[Text, Any]):
     return {key: val for key, val in args.items() if not has_tensor(val)}
 
 
@@ -65,5 +65,5 @@ def instance_class(module_class, args: Dict[Text, Any]):
     valid_keys = inspect.signature(module_class).parameters.keys()
     new_args = filter_valid_args(args, valid_keys)
     logger.info(
-        f"Instance: {module_class.__name__} {filter_out_tenosr(new_args)}.")
+        f"Instance: {module_class.__name__} {filter_out_tensor(new_args)}.")
     return module_class(**new_args)