insta-maker-3 / app.py
hivecorp's picture
Update app.py
7c14ea7 verified
import gradio as gr
import edge_tts
import asyncio
import tempfile
import nltk
import os
import srt
from pydub import AudioSegment, silence
import datetime
import nest_asyncio
nltk.download("punkt")
nest_asyncio.apply()
async def text_to_speech(text, voice, rate, pitch):
if not text.strip():
return None, None, "Please enter some text."
if not voice:
return None, None, "Please select a voice."
voice_short = voice.split(" - ")[0]
rate_str = f"{rate:+d}%"
pitch_str = f"{pitch:+d}Hz"
communicate = edge_tts.Communicate(text, voice_short, rate=rate_str, pitch=pitch_str)
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file:
tmp_path = tmp_file.name
await communicate.save(tmp_path)
srt_path = generate_srt(tmp_path, text)
return tmp_path, srt_path, ""
def generate_srt(audio_path, text):
audio = AudioSegment.from_file(audio_path)
silences = silence.detect_silence(audio, min_silence_len=400, silence_thresh=audio.dBFS - 16)
silences = [(start / 1000.0, end / 1000.0) for start, end in silences]
sentences = nltk.tokenize.sent_tokenize(text)
subtitles = []
last_time = 0.0
for i, sentence in enumerate(sentences):
if i < len(silences):
start = last_time
end = silences[i][0]
last_time = silences[i][1]
else:
start = last_time
end = start + 2.5
subtitles.append(srt.Subtitle(
index=i + 1,
start=datetime.timedelta(seconds=start),
end=datetime.timedelta(seconds=end),
content=sentence
))
srt_data = srt.compose(subtitles)
with tempfile.NamedTemporaryFile(delete=False, suffix=".srt", mode="w") as srt_file:
srt_file.write(srt_data)
return srt_file.name
async def tts_interface(text, voice, rate, pitch):
return await text_to_speech(text, voice, rate, pitch)
# ⬇️ Create demo synchronously (run async functions in loop)
voices = asyncio.run(edge_tts.list_voices())
voice_dict = {f"{v['ShortName']} - {v['Locale']} ({v['Gender']})": v['ShortName'] for v in voices}
with gr.Blocks() as demo:
gr.Markdown("# 🎙️ Text-to-Speech + Subtitle Generator")
with gr.Row():
with gr.Column():
text_input = gr.Textbox(label="Input Text", lines=5)
voice_dropdown = gr.Dropdown(choices=[""] + list(voice_dict.keys()), label="Select Voice")
rate_slider = gr.Slider(minimum=-50, maximum=50, value=0, label="Speech Rate (%)")
pitch_slider = gr.Slider(minimum=-20, maximum=20, value=0, label="Pitch (Hz)")
generate_btn = gr.Button("🎧 Generate Audio + SRT")
with gr.Column():
audio_output = gr.Audio(label="Generated Audio", type="filepath")
srt_output = gr.File(label="Download Subtitle (.srt)")
message_output = gr.Textbox(label="Status", interactive=False)
generate_btn.click(
fn=lambda text, voice, rate, pitch: asyncio.run(tts_interface(text, voice, rate, pitch)),
inputs=[text_input, voice_dropdown, rate_slider, pitch_slider],
outputs=[audio_output, srt_output, message_output]
)