Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -15,7 +15,26 @@ device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
|
15 |
print(f"Using device: {device}")
|
16 |
|
17 |
# TTS model setup
|
18 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
def download_audio_file(url: str) -> str:
|
21 |
try:
|
@@ -40,51 +59,31 @@ def extract_zip_file(zip_file: str) -> bool:
|
|
40 |
print(f"Error extracting zip file: {e}")
|
41 |
return False
|
42 |
|
43 |
-
def
|
44 |
-
|
45 |
-
if
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
tts.tts_to_file(text=text, speaker_wav=input_audio_file, language=language, file_path="./output.wav")
|
55 |
-
return "./output.wav"
|
56 |
-
|
57 |
-
@audio_processing
|
58 |
-
def process_audio(input_audio_file: str) -> str:
|
59 |
-
return input_audio_file
|
60 |
-
|
61 |
-
def voice_clone(text: str, input_file: gr.File, language: str, url: gr.Audio) -> str:
|
62 |
-
if input_file is not None:
|
63 |
-
if input_file.name.endswith(".zip"):
|
64 |
-
if extract_zip_file(input_file):
|
65 |
-
input_audio_file = [f for f in os.listdir('.') if os.path.isfile(f) and f.endswith(tuple(AUDIO_FORMATS))]
|
66 |
-
if len(input_audio_file) == 1:
|
67 |
-
input_audio_file = input_audio_file[0]
|
68 |
-
else:
|
69 |
-
return "Error: Please select a single audio file from the extracted files."
|
70 |
-
else:
|
71 |
-
input_audio_file = input_file.name
|
72 |
-
elif url:
|
73 |
-
input_audio_file = url.name
|
74 |
else:
|
75 |
-
|
76 |
-
|
77 |
-
return "Error: Invalid language selected"
|
78 |
-
output_file_path = synthesize_text(text, input_audio_file,language)
|
79 |
return output_file_path
|
80 |
|
81 |
iface = gr.Interface(
|
82 |
-
fn=
|
83 |
-
inputs=["text", gr.File(label="Input File", file_types=[".zip", *AUDIO_FORMATS]), gr.Dropdown(choices=LANGUAGES, label="Language")
|
84 |
outputs=gr.Audio(type='filepath'),
|
85 |
title='Voice Clone',
|
86 |
-
description=""" by [Angetyde](https://youtube.com/@Angetyde?si=7nusP31nTumIkPTF) and [Tony Assi](https://www.tonyassi.com/ ) use this colab with caution <3. """,
|
87 |
-
|
|
|
|
|
88 |
)
|
89 |
|
90 |
iface.launch()
|
|
|
15 |
print(f"Using device: {device}")
|
16 |
|
17 |
# TTS model setup
|
18 |
+
MODEL_PATH = "tts_models/multilingual/multi-dataset/xtts_v2"
|
19 |
+
tts = TTS(MODEL_PATH).to(device)
|
20 |
+
|
21 |
+
class AudioProcessor:
|
22 |
+
def __init__(self):
|
23 |
+
pass
|
24 |
+
|
25 |
+
def convert_to_wav(self, input_audio_file: str) -> str:
|
26 |
+
file_extension = os.path.splitext(input_audio_file)[-1].lower()
|
27 |
+
if file_extension!= ".wav":
|
28 |
+
audio = AudioSegment.from_file(input_audio_file)
|
29 |
+
audio.export("temp.wav", format="wav")
|
30 |
+
os.remove(input_audio_file)
|
31 |
+
return "temp.wav"
|
32 |
+
return input_audio_file
|
33 |
+
|
34 |
+
def synthesize_text(self, text: str, input_audio_file: str, language: str) -> str:
|
35 |
+
input_audio_file = self.convert_to_wav(input_audio_file)
|
36 |
+
tts.tts_to_file(text=text, speaker_wav=input_audio_file, language=language, file_path="./output.wav")
|
37 |
+
return "./output.wav"
|
38 |
|
39 |
def download_audio_file(url: str) -> str:
|
40 |
try:
|
|
|
59 |
print(f"Error extracting zip file: {e}")
|
60 |
return False
|
61 |
|
62 |
+
def synthesize_audio(text: str, input_file: gr.File, language: str) -> str:
|
63 |
+
audio_processor = AudioProcessor()
|
64 |
+
if input_file is None:
|
65 |
+
return None
|
66 |
+
if input_file.name.endswith(".zip"):
|
67 |
+
if extract_zip_file(input_file):
|
68 |
+
input_audio_file = [f for f in os.listdir('.') if os.path.isfile(f) and f.endswith(tuple(AUDIO_FORMATS))]
|
69 |
+
if len(input_audio_file) == 1:
|
70 |
+
input_audio_file = input_audio_file[0]
|
71 |
+
else:
|
72 |
+
return "Error: Please select a single audio file from the extracted files."
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
73 |
else:
|
74 |
+
input_audio_file = input_file.name
|
75 |
+
output_file_path = audio_processor.synthesize_text(text, input_audio_file, language)
|
|
|
|
|
76 |
return output_file_path
|
77 |
|
78 |
iface = gr.Interface(
|
79 |
+
fn=synthesize_audio,
|
80 |
+
inputs=["text", gr.File(label="Input File", file_types=[".zip", *AUDIO_FORMATS]), gr.Dropdown(choices=LANGUAGES, label="Language")],
|
81 |
outputs=gr.Audio(type='filepath'),
|
82 |
title='Voice Clone',
|
83 |
+
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.""",
|
84 |
+
examples=[["Hello! My name is Voice Clone. What is your name?", None, "en"]],
|
85 |
+
height=600,
|
86 |
+
width=1200,
|
87 |
)
|
88 |
|
89 |
iface.launch()
|