Update app.py

app.py

@@ -746,84 +746,14 @@ def generate_audio_elevenlabs(text):
         return None
 
 
-# from parler_tts import ParlerTTSForConditionalGeneration, ParlerTTSStreamer
-# from transformers import AutoTokenizer
-# from threading import Thread
-
-# repo_id = "parler-tts/parler-tts-mini-v1"
-
-
-
-
-# def generate_audio_parler_tts(text):
-#     description = "A female speaker delivers a slightly expressive and animated speech with a moderate speed and pitch. The recording is of very high quality, with the speaker's voice sounding clear and very close up."
-#     chunk_size_in_s = 0.5
-
-#     # Initialize the tokenizer and model
-#     parler_tokenizer = AutoTokenizer.from_pretrained(repo_id)
-#     parler_model = ParlerTTSForConditionalGeneration.from_pretrained(repo_id).to(device)
-#     sampling_rate = parler_model.audio_encoder.config.sampling_rate
-#     frame_rate = parler_model.audio_encoder.config.frame_rate
-
-#     def generate(text, description, play_steps_in_s=0.5):
-#         play_steps = int(frame_rate * play_steps_in_s)
-#         streamer = ParlerTTSStreamer(parler_model, device=device, play_steps=play_steps)
-
-#         inputs = parler_tokenizer(description, return_tensors="pt").to(device)
-#         prompt = parler_tokenizer(text, return_tensors="pt").to(device)
-
-#         generation_kwargs = dict(
-#             input_ids=inputs.input_ids,
-#             prompt_input_ids=prompt.input_ids,
-#             attention_mask=inputs.attention_mask,
-#             prompt_attention_mask=prompt.attention_mask,
-#             streamer=streamer,
-#             do_sample=True,
-#             temperature=1.0,
-#             min_new_tokens=10,
-#         )
-
-#         thread = Thread(target=parler_model.generate, kwargs=generation_kwargs)
-#         thread.start()
-
-#         for new_audio in streamer:
-#             if new_audio.shape[0] == 0:
-#                 break
-#             # Save or process each audio chunk as it is generated
-#             yield sampling_rate, new_audio
-
-#     audio_segments = []
-#     for (sampling_rate, audio_chunk) in generate(text, description, chunk_size_in_s):
-#         audio_segments.append(audio_chunk)
-            
-#         temp_audio_path = os.path.join(tempfile.gettempdir(), f"parler_tts_audio_chunk_{len(audio_segments)}.wav")
-#         write_wav(temp_audio_path, sampling_rate, audio_chunk.astype(np.float32))
-#         logging.debug(f"Saved chunk to {temp_audio_path}")
-        
-
-#     # Combine all the audio chunks into one audio file
-#     combined_audio = np.concatenate(audio_segments)
-#     combined_audio_path = os.path.join(tempfile.gettempdir(), "parler_tts_combined_audio_stream.wav")
-
-#     write_wav(combined_audio_path, sampling_rate, combined_audio.astype(np.float32))
-
-#     logging.debug(f"Combined audio saved to {combined_audio_path}")
-#     return combined_audio_path
-
-
-import concurrent.futures
-import tempfile
-import os
-import numpy as np
-import logging
-from queue import Queue
-from threading import Thread
-from scipy.io.wavfile import write as write_wav
 from parler_tts import ParlerTTSForConditionalGeneration, ParlerTTSStreamer
 from transformers import AutoTokenizer
+from threading import Thread
 
 repo_id = "parler-tts/parler-tts-mini-v1"
-device = "cuda:0"  # or "cpu" if CUDA is not available
+
+
+
 
 def generate_audio_parler_tts(text):
     description = "A female speaker delivers a slightly expressive and animated speech with a moderate speed and pitch. The recording is of very high quality, with the speaker's voice sounding clear and very close up."
@@ -859,39 +789,23 @@ def generate_audio_parler_tts(text):
         for new_audio in streamer:
             if new_audio.shape[0] == 0:
                 break
+            # Save or process each audio chunk as it is generated
             yield sampling_rate, new_audio
 
-    # Queue to hold the audio chunks
-    audio_queue = Queue()
-    combined_audio = []
-
-    def process_chunks():
-        while True:
-            sampling_rate, audio_chunk = audio_queue.get()
-            if audio_chunk is None:  # Stop processing when a None is received
-                break
-            combined_audio.append(audio_chunk)
-            temp_audio_path = os.path.join(tempfile.gettempdir(), f"parler_tts_audio_chunk_{len(combined_audio)}.wav")
-            write_wav(temp_audio_path, sampling_rate, audio_chunk.astype(np.float32))
-            logging.debug(f"Saved chunk to {temp_audio_path}")
-            # Start playing or buffering the audio chunk here if required
-            # (e.g., send to a player or a frontend for immediate playback)
-
-    # Start the chunk processing in a separate thread
-    with concurrent.futures.ThreadPoolExecutor() as executor:
-        executor.submit(process_chunks)
-
-        for (sampling_rate, audio_chunk) in generate(text, description, chunk_size_in_s):
-            audio_queue.put((sampling_rate, audio_chunk))
-
-    # Signal the end of processing
-    audio_queue.put((None, None))
+    audio_segments = []
+    for (sampling_rate, audio_chunk) in generate(text, description, chunk_size_in_s):
+        audio_segments.append(audio_chunk)
+            
+        temp_audio_path = os.path.join(tempfile.gettempdir(), f"parler_tts_audio_chunk_{len(audio_segments)}.wav")
+        write_wav(temp_audio_path, sampling_rate, audio_chunk.astype(np.float32))
+        logging.debug(f"Saved chunk to {temp_audio_path}")
+        
 
     # Combine all the audio chunks into one audio file
-    combined_audio_np = np.concatenate(combined_audio)
+    combined_audio = np.concatenate(audio_segments)
     combined_audio_path = os.path.join(tempfile.gettempdir(), "parler_tts_combined_audio_stream.wav")
 
-    write_wav(combined_audio_path, sampling_rate, combined_audio_np.astype(np.float32))
+    write_wav(combined_audio_path, sampling_rate, combined_audio.astype(np.float32))
 
     logging.debug(f"Combined audio saved to {combined_audio_path}")
     return combined_audio_path
@@ -901,6 +815,8 @@ def generate_audio_parler_tts(text):
 
 
 
+
+
 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}'