Update app.py

app.py

@@ -67,7 +67,7 @@ def fetch_local_events():
     api_key = os.environ['SERP_API']
     url = f'https://serpapi.com/search.json?engine=google_events&q=Events+in+Birmingham&hl=en&gl=us&api_key={api_key}'
     response = requests.get(url)
-    if response.status_code == 200:
+    if response.status_code == 200):
         events_results = response.json().get("events_results", [])
         events_html = """
         <h2 style="font-family: 'Georgia', serif; color: #ff0000; background-color: #f8f8f8; padding: 10px; border-radius: 10px;">Local Events</h2>
@@ -324,7 +324,7 @@ def fetch_local_news():
     api_key = os.environ['SERP_API']
     url = f'https://serpapi.com/search.json?engine=google_news&q=birmingham headline&api_key={api_key}'
     response = requests.get(url)
-    if response.status_code == 200:
+    if response.status_code == 200):
         results = response.json().get("news_results", [])
         news_html = """
         <h2 style="font-family: 'Georgia', serif; color: #ff0000; background-color: #f8f8f8; padding: 10px; border-radius: 10px;">Birmingham Today</h2>
@@ -547,16 +547,6 @@ def generate_audio_parler_tts(text):
 
 # Load the MARS5 model
 mars5, config_class = torch.hub.load('Camb-ai/mars5-tts', 'mars5_english', trust_repo=True)
-asr_model = pipeline(
-    "automatic-speech-recognition",
-    model="openai/whisper-tiny",
-    chunk_length_s=30,
-    device=torch.device("cuda:0"),
-)
-
-def transcribe_file(f: str) -> str:
-    predictions = asr_model(f, return_timestamps=True)["chunks"]
-    return " ".join([prediction["text"] for prediction in predictions])
 
 def generate_audio_mars5(text):
     description = "Thomas speaks with emphasis and excitement at a moderate pace with high quality."
@@ -575,12 +565,11 @@ def generate_audio_mars5(text):
 
     chunks = chunk_text(preprocess(text))
     audio_segments = []
+
     for chunk in chunks:
-        transcript = transcribe_file(audio_path)  # Assuming audio_path is the path to the audio file for reference
-        wav, sr = librosa.load(audio_path, sr=mars5.sr, mono=True)
-        wav = torch.from_numpy(wav)
+        wav = torch.zeros(1, mars5.sr)  # Use a placeholder silent audio for the reference
         cfg = config_class(**kwargs_dict)
-        ar_codes, wav_out = mars5.tts(chunk, wav, transcript.strip(), cfg=cfg)
+        ar_codes, wav_out = mars5.tts(chunk, wav, "", cfg=cfg)
         
         temp_audio_path = os.path.join(tempfile.gettempdir(), f"mars5_audio_{len(audio_segments)}.wav")
         torchaudio.save(temp_audio_path, wav_out.unsqueeze(0), mars5.sr)
@@ -651,6 +640,7 @@ demo.launch(share=True)
 
 
 
+
 # import gradio as gr
 # import requests
 # import os