Update app.py

app.py

@@ -949,76 +949,76 @@ def generate_audio_elevenlabs(text):
 
 # chunking audio and then Process
 
-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
+# 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
+
+# # Ensure your device is set to CUDA
+# device = "cuda:0" if torch.cuda.is_available() else "cpu"
 
-# Ensure your device is set to CUDA
-device = "cuda:0" if torch.cuda.is_available() else "cpu"
+# repo_id = "parler-tts/parler-tts-mini-v1"
 
-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,
-        )
+# 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()
+#         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
+#         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)
+#     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}")
+#         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")
+#     # 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))
+#     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
+#     logging.debug(f"Combined audio saved to {combined_audio_path}")
+#     return combined_audio_path
 
 
 def fetch_local_events():