Rename voice_infer_xtts.py to bark_infer_cpu.py

bark_infer_cpu.py

@@ -0,0 +1,22 @@
+from bark import generate_audio, preload_models
+import numpy as np
+import scipy.io.wavfile
+import argparse
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--text_file", required=True)
+parser.add_argument("--output", default="bark_output.wav")
+args = parser.parse_args()
+
+# Read text
+with open(args.text_file, "r") as f:
+    text = f.read()
+
+# Load Bark CPU models
+preload_models(use_gpu=False)
+
+# Generate audio
+audio_array = generate_audio(text, history_prompt="v2/en_speaker_9")
+
+# Save output
+scipy.io.wavfile.write(args.output, rate=22050, data=audio_array)

voice_infer_xtts.py

@@ -1,22 +0,0 @@
-# voice_infer_xtts.py
-
-import torch
-from TTS.api import TTS
-import argparse
-
-parser = argparse.ArgumentParser()
-parser.add_argument("--text_file", required=True)
-parser.add_argument("--output", default="xtts_output.wav")
-parser.add_argument("--language", default="hi")  # 'hi' for Hindi, 'en' for English
-parser.add_argument("--voice", default="female")  # Optionally switch voice profiles
-args = parser.parse_args()
-
-# Read text
-with open(args.text_file, "r") as f:
-    text = f.read()
-
-# Load Coqui XTTS
-tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", gpu=True)
-
-# Generate voice
-tts.tts_to_file(text=text, file_path=args.output, language=args.language, speaker=args.voice)