Spaces:
Running
Running
| import gradio as gr | |
| import os | |
| import requests | |
| import torch | |
| import zipfile | |
| from TTS.api import TTS | |
| from pydub import AudioSegment | |
| # Constants | |
| AUDIO_FORMATS = [".wav", ".mp3", ".flac", ".mp4"] | |
| LANGUAGES = ["en", "es", "fr", "de", "it", "pt", "pl", "tr", "ru", "nl", "cs", "ar", "zh-cn", "ja", "hu", "ko", "hi"] | |
| # Device setup | |
| device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") | |
| print(f"Using device: {device}") | |
| # TTS model setup | |
| 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: | |
| response = requests.get(url) | |
| file_extension = os.path.splitext(url)[-1].lower() | |
| if file_extension not in AUDIO_FORMATS: | |
| raise ValueError(f"Unsupported file extension: {file_extension}") | |
| file_name = f"temp{file_extension}" | |
| with open(file_name, "wb") as f: | |
| f.write(response.content) | |
| return file_name | |
| except requests.exceptions.RequestException as e: | |
| print(f"Error downloading audio file: {e}") | |
| return None | |
| def extract_zip_file(zip_file: str) -> bool: | |
| try: | |
| with zipfile.ZipFile(zip_file, 'r') as zip_ref: | |
| zip_ref.extractall() | |
| return True | |
| except zipfile.BadZipfile as e: | |
| print(f"Error extracting zip file: {e}") | |
| return False | |
| 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: | |
| 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=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. 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() |