Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -6,10 +6,7 @@ from dotenv import load_dotenv
|
|
6 |
|
7 |
load_dotenv()
|
8 |
|
9 |
-
def
|
10 |
-
return asyncio.run(synthesize(article_url, text_input, language))
|
11 |
-
|
12 |
-
async def synthesize(article_url, text_input, language):
|
13 |
if not article_url and not text_input:
|
14 |
return "Error: Ingresa una URL o texto", None
|
15 |
|
@@ -17,7 +14,7 @@ async def synthesize(article_url, text_input, language):
|
|
17 |
config = ConversationConfig()
|
18 |
converter = URLToAudioConverter(config, llm_api_key=os.environ.get("TOGETHER_API_KEY"))
|
19 |
|
20 |
-
# Voces
|
21 |
voices = {
|
22 |
"en": ("en-US-AvaMultilingualNeural", "en-US-AndrewMultilingualNeural"),
|
23 |
"es": ("es-ES-AlvaroNeural", "es-ES-ElviraNeural")
|
@@ -25,34 +22,28 @@ async def synthesize(article_url, text_input, language):
|
|
25 |
voice1, voice2 = voices.get(language, voices["en"])
|
26 |
|
27 |
if text_input:
|
28 |
-
|
29 |
-
conversation_json = converter.extract_conversation(text_input)
|
30 |
-
conversation_text = "\n".join(
|
31 |
-
f"{turn['speaker']}: {turn['text']}"
|
32 |
-
for turn in conversation_json["conversation"]
|
33 |
-
)
|
34 |
-
audio_files, _ = await converter.text_to_speech(conversation_json, voice1, voice2)
|
35 |
-
output_file = os.path.join("combined_output.wav")
|
36 |
-
converter.combine_audio_files(audio_files, output_file)
|
37 |
else:
|
38 |
-
|
39 |
-
output_file, conversation_text = await converter.url_to_audio(article_url, voice1, voice2)
|
40 |
|
41 |
-
return
|
42 |
except Exception as e:
|
43 |
return f"Error: {str(e)}", None
|
44 |
|
|
|
|
|
|
|
45 |
with gr.Blocks(theme='gstaff/sketch') as demo:
|
46 |
gr.Markdown("# 🎙 Podcast Converter (Human Voices)")
|
47 |
with gr.Group():
|
48 |
-
text_url = gr.Textbox(label="
|
49 |
-
text_input = gr.Textbox(label="
|
50 |
-
language = gr.Dropdown(["en", "es"], label="
|
51 |
-
btn = gr.Button("
|
52 |
|
53 |
with gr.Row():
|
54 |
-
conv_display = gr.Textbox(label="
|
55 |
-
aud = gr.Audio(label="
|
56 |
|
57 |
btn.click(
|
58 |
synthesize_sync,
|
|
|
6 |
|
7 |
load_dotenv()
|
8 |
|
9 |
+
async def synthesize(article_url, text_input, language="en"):
|
|
|
|
|
|
|
10 |
if not article_url and not text_input:
|
11 |
return "Error: Ingresa una URL o texto", None
|
12 |
|
|
|
14 |
config = ConversationConfig()
|
15 |
converter = URLToAudioConverter(config, llm_api_key=os.environ.get("TOGETHER_API_KEY"))
|
16 |
|
17 |
+
# Voces humanizadas
|
18 |
voices = {
|
19 |
"en": ("en-US-AvaMultilingualNeural", "en-US-AndrewMultilingualNeural"),
|
20 |
"es": ("es-ES-AlvaroNeural", "es-ES-ElviraNeural")
|
|
|
22 |
voice1, voice2 = voices.get(language, voices["en"])
|
23 |
|
24 |
if text_input:
|
25 |
+
output_file, conversation = await converter.text_to_audio(text_input, voice1, voice2)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
else:
|
27 |
+
output_file, conversation = await converter.url_to_audio(article_url, voice1, voice2)
|
|
|
28 |
|
29 |
+
return conversation, output_file
|
30 |
except Exception as e:
|
31 |
return f"Error: {str(e)}", None
|
32 |
|
33 |
+
def synthesize_sync(article_url, text_input, language):
|
34 |
+
return asyncio.run(synthesize(article_url, text_input, language))
|
35 |
+
|
36 |
with gr.Blocks(theme='gstaff/sketch') as demo:
|
37 |
gr.Markdown("# 🎙 Podcast Converter (Human Voices)")
|
38 |
with gr.Group():
|
39 |
+
text_url = gr.Textbox(label="URL (opcional)", placeholder="https://...")
|
40 |
+
text_input = gr.Textbox(label="O texto directo", lines=5)
|
41 |
+
language = gr.Dropdown(["en", "es"], label="Idioma", value="en")
|
42 |
+
btn = gr.Button("Generar Podcast", variant="primary")
|
43 |
|
44 |
with gr.Row():
|
45 |
+
conv_display = gr.Textbox(label="Conversación", interactive=False, lines=10)
|
46 |
+
aud = gr.Audio(label="Audio Generado", interactive=False)
|
47 |
|
48 |
btn.click(
|
49 |
synthesize_sync,
|