<add> Chinese mandarin implementation (tacotron2)

TTS/tts/datasets/preprocess.py

@@ -372,15 +372,10 @@ def _voxcel_x(root_path, meta_file, voxcel_idx):
 
 
 
-def baker(root_path: str, meta_file: str) ->  List[List[str]]:
-    """Normalizes the Baker meta data file to TTS format
 
-    Args:
-        root_path (str): path to the baker dataset
-        meta_file (str): name of the meta dataset containing names of wav to select and the transcript of the sentence  
-    Returns:
-        List[List[str]]: List of (text, wav_path, speaker_name) associated with each sentences
-    """
+# ======================================== Baker (chinese mandarin single speaker) ===========================================
+def baker(root_path, meta_file):
+    """Normalizes the Baker meta data file to TTS format"""
     txt_file = os.path.join(root_path, meta_file)
     items = []
     speaker_name = "baker"
@@ -389,4 +384,4 @@ def baker(root_path: str, meta_file: str) ->  List[List[str]]:
             wav_name, text = line.rstrip('\n').split("|")
             wav_path = os.path.join(root_path, "clips_22", wav_name)
             items.append([text, wav_path, speaker_name])
-    return items 
+    return items

TTS/tts/utils/synthesis.py

@@ -1,5 +1,5 @@
 import os
-os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' 
+os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
 import pkg_resources
 installed = {pkg.key for pkg in pkg_resources.working_set}  #pylint: disable=not-an-iterable
 if 'tensorflow' in installed or 'tensorflow-gpu' in installed:
@@ -220,6 +220,7 @@ def synthesis(model,
                 model outputs.
             speaker_id (int): id of speaker
             style_wav (str | Dict[str, float]): Uses for style embedding of GST.
+            style_wav (str): Uses for style embedding of GST.
             truncated (bool): keep model states after inference. It can be used
                 for continuous inference at long texts.
             enable_eos_bos_chars (bool): enable special chars for end of sentence and start of sentence.

TTS/utils/synthesizer.py

@@ -122,6 +122,13 @@ class Synthesizer(object):
         speaker_embedding = self.init_speaker(speaker_idx)
         use_gl = self.vocoder_model is None
 
+
+        # check if compute gst style
+        gst_style_input = None
+        if self.tts_config.use_gst:
+            if self.tts_config.gst["gst_style_input"] not in ["", {}]:
+                style_wav = self.tts_config.gst["gst_style_input"]
+
         for sen in sens:
             # synthesize voice
             waveform, _, _, mel_postnet_spec, _, _ = synthesis(
@@ -131,7 +138,7 @@ class Synthesizer(object):
                 self.use_cuda,
                 self.ap,
                 speaker_idx,
-                None,
+                gst_style_input,
                 False,
                 self.tts_config.enable_eos_bos_chars,
                 use_gl,