Spaces:
Sleeping
Sleeping
Upload streamlit_app.py
Browse files- streamlit_app.py +143 -1
streamlit_app.py
CHANGED
|
@@ -1,2 +1,144 @@
|
|
| 1 |
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
|
| 2 |
+
import streamlit as st
|
| 3 |
+
import tempfile
|
| 4 |
+
from faster_whisper import WhisperModel
|
| 5 |
+
import textwrap
|
| 6 |
+
from datetime import timedelta
|
| 7 |
+
from xml.sax.saxutils import escape
|
| 8 |
+
|
| 9 |
+
st.set_page_config(page_title="VO to Subtitle Generator Multi-languages version", layout="wide")
|
| 10 |
+
st.title("🌍 VO to Subtitle Generator — Multi-languages Version")
|
| 11 |
+
|
| 12 |
+
def format_time(seconds):
|
| 13 |
+
td = timedelta(seconds=seconds)
|
| 14 |
+
result = str(td)[:11].replace(".", ",")
|
| 15 |
+
return result if "," in result else result + ",000"
|
| 16 |
+
|
| 17 |
+
def generate_srt(segments, max_chars, max_lines):
|
| 18 |
+
srt_text = ""
|
| 19 |
+
count = 1
|
| 20 |
+
for seg in segments:
|
| 21 |
+
start = seg.start
|
| 22 |
+
end = seg.end
|
| 23 |
+
text = seg.text.strip()
|
| 24 |
+
lines = textwrap.wrap(text, width=max_chars)
|
| 25 |
+
grouped = [lines[i:i+max_lines] for i in range(0, len(lines), max_lines)]
|
| 26 |
+
chunk_count = len(grouped)
|
| 27 |
+
duration = end - start
|
| 28 |
+
chunk_duration = duration / chunk_count if chunk_count > 0 else duration
|
| 29 |
+
|
| 30 |
+
for j, chunk in enumerate(grouped):
|
| 31 |
+
chunk_start = start + j * chunk_duration
|
| 32 |
+
chunk_end = chunk_start + chunk_duration
|
| 33 |
+
timestamp = f"{format_time(chunk_start)} --> {format_time(chunk_end)}"
|
| 34 |
+
content = "\n".join(chunk)
|
| 35 |
+
srt_text += f"{count}\n{timestamp}\n{content}\n\n"
|
| 36 |
+
count += 1
|
| 37 |
+
return srt_text
|
| 38 |
+
|
| 39 |
+
def generate_fcpxml(segments, version):
|
| 40 |
+
xml = [f'<?xml version="1.0" encoding="UTF-8"?>',
|
| 41 |
+
f'<!DOCTYPE fcpxml>',
|
| 42 |
+
f'<fcpxml version="{version}">',
|
| 43 |
+
' <resources>',
|
| 44 |
+
' <format id="r1" name="FFVideoFormat1080p30" frameDuration="100/3000s" width="1920" height="1080"/>',
|
| 45 |
+
' <effect id="r2" name="Basic Title" uid=".../Effects.localized/Basic Titles.localized/Basic Title.localized/Basic Title.moti"/>',
|
| 46 |
+
' </resources>',
|
| 47 |
+
' <library>',
|
| 48 |
+
' <event name="Transcription">',
|
| 49 |
+
' <project name="Subtitles">',
|
| 50 |
+
f' <sequence duration="{round(segments[-1].end, 2)}s" format="r1" tcStart="0s" tcFormat="NDF">',
|
| 51 |
+
' <spine>']
|
| 52 |
+
for seg in segments:
|
| 53 |
+
start = round(seg.start, 2)
|
| 54 |
+
duration = round(seg.end - seg.start, 2)
|
| 55 |
+
text = escape(seg.text.strip())
|
| 56 |
+
xml.append(
|
| 57 |
+
f' <title lane="1" offset="{start}s" ref="r2" duration="{duration}s" start="{start}s">'
|
| 58 |
+
f' <text>{text}</text>'
|
| 59 |
+
f' </title>'
|
| 60 |
+
)
|
| 61 |
+
xml.extend([' </spine>',
|
| 62 |
+
' </sequence>',
|
| 63 |
+
' </project>',
|
| 64 |
+
' </event>',
|
| 65 |
+
' </library>',
|
| 66 |
+
'</fcpxml>'])
|
| 67 |
+
return '\n'.join(xml)
|
| 68 |
+
|
| 69 |
+
with st.sidebar:
|
| 70 |
+
st.header("⚙️ Settings")
|
| 71 |
+
uploaded_file = st.file_uploader("Upload MP3 or WAV", type=["mp3", "wav"])
|
| 72 |
+
model_size = st.selectbox("Model Size", ["tiny", "base", "small", "medium"])
|
| 73 |
+
layout = st.selectbox("Video Layout", ["Horizontal (37 chars)", "Vertical (25 chars)"])
|
| 74 |
+
lines = st.selectbox("Lines per Subtitle", [1, 2], index=1)
|
| 75 |
+
language_map = {
|
| 76 |
+
"Auto": None,
|
| 77 |
+
"Arabic": "ar",
|
| 78 |
+
"English": "en",
|
| 79 |
+
"French": "fr",
|
| 80 |
+
"Farsi": "fa",
|
| 81 |
+
"Spanish": "es"
|
| 82 |
+
}
|
| 83 |
+
language = st.selectbox("Language", list(language_map.keys()))
|
| 84 |
+
export_format = st.selectbox("Export Format", ["srt", "fcpxml"])
|
| 85 |
+
fcpxml_version = st.selectbox("FCPXML Version", ["1.13", "1.12", "1.11"], index=0) if export_format == "fcpxml" else None
|
| 86 |
+
|
| 87 |
+
if 'subtitle_data' not in st.session_state:
|
| 88 |
+
st.session_state.subtitle_data = ""
|
| 89 |
+
st.session_state.text_dir = "rtl"
|
| 90 |
+
st.session_state.generated = False
|
| 91 |
+
|
| 92 |
+
if uploaded_file and st.button("🔁 Generate Subtitle"):
|
| 93 |
+
with st.spinner("Transcribing with Whisper..."):
|
| 94 |
+
with tempfile.NamedTemporaryFile(delete=False) as temp_audio:
|
| 95 |
+
temp_audio.write(uploaded_file.read())
|
| 96 |
+
temp_audio.flush()
|
| 97 |
+
whisper = WhisperModel(model_size, device="cpu", compute_type="int8")
|
| 98 |
+
segments_gen, _ = whisper.transcribe(temp_audio.name, language=language_map[language])
|
| 99 |
+
segments = list(segments_gen)
|
| 100 |
+
st.session_state.segments = segments
|
| 101 |
+
|
| 102 |
+
max_chars = 25 if "Vertical" in layout else 37
|
| 103 |
+
max_lines = int(lines)
|
| 104 |
+
|
| 105 |
+
if export_format == "srt":
|
| 106 |
+
st.session_state.subtitle_data = generate_srt(segments, max_chars, max_lines)
|
| 107 |
+
else:
|
| 108 |
+
st.session_state.subtitle_data = generate_fcpxml(segments, fcpxml_version)
|
| 109 |
+
|
| 110 |
+
st.session_state.generated = True
|
| 111 |
+
|
| 112 |
+
if st.session_state.generated and st.session_state.subtitle_data:
|
| 113 |
+
col1, col2 = st.columns([1, 6])
|
| 114 |
+
with col1:
|
| 115 |
+
st.write("Text Direction:")
|
| 116 |
+
if st.button("⬅️ RTL"):
|
| 117 |
+
st.session_state.text_dir = "rtl"
|
| 118 |
+
if st.button("➡️ LTR"):
|
| 119 |
+
st.session_state.text_dir = "ltr"
|
| 120 |
+
|
| 121 |
+
with col2:
|
| 122 |
+
st.markdown("### ✏️ Edit Before Download")
|
| 123 |
+
preview_html = f'''
|
| 124 |
+
<textarea id="subtitle_editor" name="subtitle_editor"
|
| 125 |
+
style="width: 100%; height: 300px; padding: 1.5em 2em;
|
| 126 |
+
border: 1px solid #ccc; border-radius: 8px;
|
| 127 |
+
font-family: monospace; font-size: 14px;
|
| 128 |
+
direction: {st.session_state.text_dir}; white-space: pre-wrap;">{st.session_state.subtitle_data}</textarea>
|
| 129 |
+
<script>
|
| 130 |
+
const editor = document.getElementById('subtitle_editor');
|
| 131 |
+
editor.addEventListener('input', () => {{
|
| 132 |
+
window.subtitleEdited = editor.value;
|
| 133 |
+
}});
|
| 134 |
+
window.subtitleEdited = editor.value;
|
| 135 |
+
</script>
|
| 136 |
+
'''
|
| 137 |
+
st.components.v1.html(preview_html, height=360)
|
| 138 |
+
|
| 139 |
+
st.download_button(
|
| 140 |
+
label="⬇️ Download Subtitle",
|
| 141 |
+
data=st.session_state.subtitle_data,
|
| 142 |
+
file_name="subtitles." + ("srt" if export_format == "srt" else "fcpxml"),
|
| 143 |
+
mime="text/plain"
|
| 144 |
+
)
|