Rajhuggingface4253 commited on
Commit
74ce059
·
verified ·
1 Parent(s): 1b19075

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +3 -9
app.py CHANGED
@@ -143,11 +143,7 @@ class NeuTTSWrapper:
143
 
144
  def generate_speech_blocking(self, text: str, ref_audio_path: str) -> np.ndarray:
145
  """Blocking synthesis for standard endpoint."""
146
- # 1. Load reference
147
- reference_audio, sr = sf.read(ref_audio_path)
148
- if sr != SAMPLE_RATE:
149
- # Simple check/resize logic required if sample rate mismatch occurs
150
- pass
151
 
152
  # 2. Encode reference
153
  ref_s = self.tts_model.encode_reference(reference_audio)
@@ -161,11 +157,9 @@ class NeuTTSWrapper:
161
  """Sentence-by-Sentence Streaming (Blocking)."""
162
  logger.info(f"Starting streaming synthesis for text length: {len(text)}")
163
 
164
- # 1. Load reference audio (ONLY ONCE)
165
- reference_audio, sr = sf.read(ref_audio_path)
166
 
167
- # 2. Encode reference (ONLY ONCE)
168
- ref_s = self.tts_model.encode_reference(reference_audio)
169
 
170
  # 3. Split text
171
  sentences = self._split_text_into_chunks(text)
 
143
 
144
  def generate_speech_blocking(self, text: str, ref_audio_path: str) -> np.ndarray:
145
  """Blocking synthesis for standard endpoint."""
146
+
 
 
 
 
147
 
148
  # 2. Encode reference
149
  ref_s = self.tts_model.encode_reference(reference_audio)
 
157
  """Sentence-by-Sentence Streaming (Blocking)."""
158
  logger.info(f"Starting streaming synthesis for text length: {len(text)}")
159
 
 
 
160
 
161
+
162
+ ref_s = self.tts_model.encode_reference(ref_audio_path)
163
 
164
  # 3. Split text
165
  sentences = self._split_text_into_chunks(text)