Update LLMwithvoice.py

LLMwithvoice.py

@@ -1,9 +1,10 @@
 import requests
 import torch
 import numpy as np
-import sounddevice as sd
 from transformers import AutoTokenizer
 from parler_tts import ParlerTTSForConditionalGeneration
+from pydub import AudioSegment
+import simpleaudio as sa
 
 # Hugging Face API URL for Roberta model
 API_URL_ROBERTA = "https://api-inference.huggingface.co/models/deepset/roberta-base-squad2"
@@ -47,10 +48,23 @@ def generate_speech(answer):
     generation = model.generate(input_ids=input_ids).to(torch.float32)
     audio_arr = generation.cpu().numpy().squeeze()
 
-    # Play the generated audio
+    # Convert numpy array to audio segment
+    audio_segment = AudioSegment(
+        audio_arr.tobytes(),
+        frame_rate=model.config.sampling_rate,
+        sample_width=audio_arr.dtype.itemsize,
+        channels=1
+    )
+
+    # Play the audio using simpleaudio
     try:
-        sd.play(audio_arr, samplerate=model.config.sampling_rate)
-        sd.wait()  # Wait until the audio is done playing
+        play_obj = sa.play_buffer(
+            audio_segment.raw_data,
+            num_channels=1,
+            bytes_per_sample=audio_segment.sample_width,
+            sample_rate=audio_segment.frame_rate
+        )
+        play_obj.wait_done()  # Wait until the audio is done playing
     except Exception as e:
         print(f"Error playing audio: {e}")