Update app.py

app.py

@@ -15,7 +15,26 @@ device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 print(f"Using device: {device}")
 
 # TTS model setup
-tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
+MODEL_PATH = "tts_models/multilingual/multi-dataset/xtts_v2"
+tts = TTS(MODEL_PATH).to(device)
+
+class AudioProcessor:
+ def __init__(self):
+ pass
+
+ def convert_to_wav(self, input_audio_file: str) -> str:
+ file_extension = os.path.splitext(input_audio_file)[-1].lower()
+ if file_extension!= ".wav":
+ audio = AudioSegment.from_file(input_audio_file)
+ audio.export("temp.wav", format="wav")
+ os.remove(input_audio_file)
+ return "temp.wav"
+ return input_audio_file
+
+ def synthesize_text(self, text: str, input_audio_file: str, language: str) -> str:
+ input_audio_file = self.convert_to_wav(input_audio_file)
+ tts.tts_to_file(text=text, speaker_wav=input_audio_file, language=language, file_path="./output.wav")
+ return "./output.wav"
 
 def download_audio_file(url: str) -> str:
  try:
@@ -40,51 +59,31 @@ def extract_zip_file(zip_file: str) -> bool:
  print(f"Error extracting zip file: {e}")
  return False
 
-def convert_to_wav(input_audio_file: str) -> str:
- file_extension = os.path.splitext(input_audio_file)[-1].lower()
- if file_extension!= ".wav":
- audio = AudioSegment.from_file(input_audio_file)
- audio.export("temp.wav", format="wav")
- os.remove(input_audio_file)
- return "temp.wav"
- return input_audio_file
-
-def synthesize_text(text: str, input_audio_file: str, language: str) -> str:
- input_audio_file = convert_to_wav(input_audio_file)
- tts.tts_to_file(text=text, speaker_wav=input_audio_file, language=language, file_path="./output.wav")
- return "./output.wav"
-
-@audio_processing
-def process_audio(input_audio_file: str) -> str:
- return input_audio_file
-
-def voice_clone(text: str, input_file: gr.File, language: str, url: gr.Audio) -> str:
- if input_file is not None:
- if input_file.name.endswith(".zip"):
- if extract_zip_file(input_file):
- input_audio_file = [f for f in os.listdir('.') if os.path.isfile(f) and f.endswith(tuple(AUDIO_FORMATS))]
- if len(input_audio_file) == 1:
- input_audio_file = input_audio_file[0]
- else:
- return "Error: Please select a single audio file from the extracted files."
- else:
- input_audio_file = input_file.name
- elif url:
- input_audio_file = url.name
+def synthesize_audio(text: str, input_file: gr.File, language: str) -> str:
+ audio_processor = AudioProcessor()
+ if input_file is None:
+ return None
+ if input_file.name.endswith(".zip"):
+ if extract_zip_file(input_file):
+ input_audio_file = [f for f in os.listdir('.') if os.path.isfile(f) and f.endswith(tuple(AUDIO_FORMATS))]
+ if len(input_audio_file) == 1:
+ input_audio_file = input_audio_file[0]
+ else:
+ return "Error: Please select a single audio file from the extracted files."
  else:
- return"Error: Please select a file or enter a URL"
- if language not in LANGUAGES:
- return "Error: Invalid language selected"
- output_file_path = synthesize_text(text, input_audio_file,language)
+ input_audio_file = input_file.name
+ output_file_path = audio_processor.synthesize_text(text, input_audio_file, language)
  return output_file_path
 
 iface = gr.Interface(
- fn=voice_clone,
- inputs=["text", gr.File(label="Input File", file_types=[".zip", *AUDIO_FORMATS]), gr.Dropdown(choices=LANGUAGES, label="Language"), gr.Audio(label="URL")],
+ fn=synthesize_audio,
+ inputs=["text", gr.File(label="Input File", file_types=[".zip", *AUDIO_FORMATS]), gr.Dropdown(choices=LANGUAGES, label="Language")],
  outputs=gr.Audio(type='filepath'),
  title='Voice Clone',
- description=""" by [Angetyde](https://youtube.com/@Angetyde?si=7nusP31nTumIkPTF) and [Tony Assi](https://www.tonyassi.com/ ) use this colab with caution <3. """,
- theme=gr.themes.Base(primary_hue="green"),
+ description=""" by [Angetyde](https://youtube.com/@Angetyde?si=7nusP31nTumIkPTF) and [Tony Assi](https://www.tonyassi.com/ ) use this colab with caution <3. Clone any voice with a model and generate a speech waveform.""",
+ examples=[["Hello! My name is Voice Clone. What is your name?", None, "en"]],
+ height=600,
+ width=1200,
 )
 
 iface.launch()