Update app.py

app.py

@@ -10,7 +10,7 @@ import numpy as np
 PODCAST_SUBJECT = "The future of AI and its impact on society"
 
 # Initialize the inference client
-client = InferenceClient("Qwen/Qwen2.5-Coder-32B-Instruct", provider="together", token=os.getenv("HF_TOKEN"))
+client = InferenceClient("meta-llama/Llama-3.3-70B-Instruct", provider="cerebras", token=os.getenv("HF_TOKEN"))
 model = Dia.from_pretrained("nari-labs/Dia-1.6B", compute_dtype="float16")
 
 # Queue for audio streaming
@@ -45,12 +45,12 @@ def split_podcast_into_chunks(podcast_text, chunk_size=3):
 
 def process_audio_chunks(podcast_text):
     chunks = split_podcast_into_chunks(podcast_text)
-    sample_rate = 36000 # Modified from https://huggingface.co/spaces/nari-labs/Dia-1.6B/blob/main/app.py has 44100
+    sample_rate = 39000 # Modified from https://huggingface.co/spaces/nari-labs/Dia-1.6B/blob/main/app.py has 44100
     for chunk in chunks:
         if stop_signal.is_set():
             break
 
-        raw_audio = model.generate(chunk, use_torch_compile=True, verbose=False)
+        raw_audio = model.generate(chunk, use_torch_compile=True, verbose=False, seed=42)
         audio_chunk = np.array(raw_audio, dtype=np.float32)
         audio_queue.put((sample_rate, audio_chunk))