Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -7,6 +7,9 @@ import re
|
|
| 7 |
from pydub import AudioSegment
|
| 8 |
from openpyxl import Workbook
|
| 9 |
from openpyxl.styles import Font
|
|
|
|
|
|
|
|
|
|
| 10 |
from io import BytesIO
|
| 11 |
|
| 12 |
# ---------------------------------------------------
|
|
@@ -29,7 +32,7 @@ model_option = st.sidebar.selectbox(
|
|
| 29 |
)
|
| 30 |
|
| 31 |
output_mode = st.sidebar.radio(
|
| 32 |
-
"Output
|
| 33 |
["Roman Urdu", "English"]
|
| 34 |
)
|
| 35 |
|
|
@@ -41,7 +44,7 @@ if st.sidebar.button("🧹 Clear Session"):
|
|
| 41 |
# HEADER
|
| 42 |
# ---------------------------------------------------
|
| 43 |
st.markdown("<h1 style='text-align:center;'>🎤 RecToText Pro</h1>", unsafe_allow_html=True)
|
| 44 |
-
st.markdown("<p style='text-align:center;'>AI
|
| 45 |
st.divider()
|
| 46 |
|
| 47 |
# ---------------------------------------------------
|
|
@@ -57,7 +60,22 @@ def clean_text(text):
|
|
| 57 |
pattern = r'\b(?:' + '|'.join(filler_words) + r')\b'
|
| 58 |
text = re.sub(pattern, '', text, flags=re.IGNORECASE)
|
| 59 |
text = re.sub(r'\s+', ' ', text).strip()
|
| 60 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
| 62 |
def convert_to_roman_urdu(text):
|
| 63 |
replacements = {
|
|
@@ -95,13 +113,31 @@ def create_excel(segments):
|
|
| 95 |
excel_buffer.seek(0)
|
| 96 |
return excel_buffer
|
| 97 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
# ---------------------------------------------------
|
| 99 |
-
# FILE UPLOADER
|
| 100 |
# ---------------------------------------------------
|
| 101 |
uploaded_file = st.file_uploader(
|
| 102 |
"Upload Lecture Recording (.mp3, .wav, .m4a, .aac)",
|
| 103 |
-
type=["mp3", "wav", "m4a", "aac"]
|
| 104 |
-
help="Supported formats: mp3, wav, m4a, aac"
|
| 105 |
)
|
| 106 |
|
| 107 |
if uploaded_file:
|
|
@@ -117,15 +153,12 @@ if uploaded_file:
|
|
| 117 |
st.info("Loading Whisper model...")
|
| 118 |
model = load_model(model_option)
|
| 119 |
|
| 120 |
-
progress = st.progress(0)
|
| 121 |
start_time = time.time()
|
| 122 |
|
| 123 |
with st.spinner("Transcribing... Please wait."):
|
| 124 |
result = model.transcribe(temp_audio_path)
|
| 125 |
-
progress.progress(100)
|
| 126 |
|
| 127 |
end_time = time.time()
|
| 128 |
-
|
| 129 |
os.remove(temp_audio_path)
|
| 130 |
|
| 131 |
detected_lang = result.get("language", "Unknown")
|
|
@@ -140,9 +173,6 @@ if uploaded_file:
|
|
| 140 |
word_count = len(cleaned_text.split())
|
| 141 |
processing_time = round(end_time - start_time, 2)
|
| 142 |
|
| 143 |
-
# ---------------------------------------------------
|
| 144 |
-
# DISPLAY RESULTS
|
| 145 |
-
# ---------------------------------------------------
|
| 146 |
col1, col2 = st.columns(2)
|
| 147 |
|
| 148 |
with col1:
|
|
@@ -150,7 +180,7 @@ if uploaded_file:
|
|
| 150 |
st.text_area("", full_text, height=350)
|
| 151 |
|
| 152 |
with col2:
|
| 153 |
-
st.subheader("✨
|
| 154 |
st.text_area("", cleaned_text, height=350)
|
| 155 |
|
| 156 |
st.divider()
|
|
@@ -159,23 +189,32 @@ if uploaded_file:
|
|
| 159 |
st.write(f"**Word Count:** {word_count}")
|
| 160 |
st.write(f"**Processing Time:** {processing_time} seconds")
|
| 161 |
|
| 162 |
-
# ----------------------------
|
| 163 |
-
#
|
| 164 |
-
# ----------------------------
|
| 165 |
excel_file = create_excel(segments)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 166 |
|
| 167 |
-
st.download_button(
|
| 168 |
-
label="📥 Download Excel File",
|
| 169 |
-
data=excel_file,
|
| 170 |
-
file_name="RecToText_Transcription.xlsx",
|
| 171 |
-
mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
|
| 172 |
-
)
|
| 173 |
-
|
| 174 |
-
# ---------------------------------------------------
|
| 175 |
-
# FOOTER
|
| 176 |
-
# ---------------------------------------------------
|
| 177 |
st.divider()
|
| 178 |
st.markdown(
|
| 179 |
-
"<p style='text-align:center; font-size:12px;'>
|
| 180 |
unsafe_allow_html=True
|
| 181 |
)
|
|
|
|
| 7 |
from pydub import AudioSegment
|
| 8 |
from openpyxl import Workbook
|
| 9 |
from openpyxl.styles import Font
|
| 10 |
+
from docx import Document
|
| 11 |
+
from docx.shared import Pt
|
| 12 |
+
from docx.enum.text import WD_ALIGN_PARAGRAPH
|
| 13 |
from io import BytesIO
|
| 14 |
|
| 15 |
# ---------------------------------------------------
|
|
|
|
| 32 |
)
|
| 33 |
|
| 34 |
output_mode = st.sidebar.radio(
|
| 35 |
+
"Output Format",
|
| 36 |
["Roman Urdu", "English"]
|
| 37 |
)
|
| 38 |
|
|
|
|
| 44 |
# HEADER
|
| 45 |
# ---------------------------------------------------
|
| 46 |
st.markdown("<h1 style='text-align:center;'>🎤 RecToText Pro</h1>", unsafe_allow_html=True)
|
| 47 |
+
st.markdown("<p style='text-align:center;'>AI Lecture Transcriber with Excel & Word Export</p>", unsafe_allow_html=True)
|
| 48 |
st.divider()
|
| 49 |
|
| 50 |
# ---------------------------------------------------
|
|
|
|
| 60 |
pattern = r'\b(?:' + '|'.join(filler_words) + r')\b'
|
| 61 |
text = re.sub(pattern, '', text, flags=re.IGNORECASE)
|
| 62 |
text = re.sub(r'\s+', ' ', text).strip()
|
| 63 |
+
|
| 64 |
+
# Better paragraph formatting
|
| 65 |
+
sentences = re.split(r'(?<=[.!?]) +', text)
|
| 66 |
+
paragraphs = []
|
| 67 |
+
temp_para = ""
|
| 68 |
+
|
| 69 |
+
for i, sentence in enumerate(sentences):
|
| 70 |
+
temp_para += sentence + " "
|
| 71 |
+
if (i + 1) % 5 == 0:
|
| 72 |
+
paragraphs.append(temp_para.strip())
|
| 73 |
+
temp_para = ""
|
| 74 |
+
|
| 75 |
+
if temp_para:
|
| 76 |
+
paragraphs.append(temp_para.strip())
|
| 77 |
+
|
| 78 |
+
return "\n\n".join(paragraphs)
|
| 79 |
|
| 80 |
def convert_to_roman_urdu(text):
|
| 81 |
replacements = {
|
|
|
|
| 113 |
excel_buffer.seek(0)
|
| 114 |
return excel_buffer
|
| 115 |
|
| 116 |
+
def create_word_document(cleaned_text):
|
| 117 |
+
doc = Document()
|
| 118 |
+
|
| 119 |
+
title = doc.add_heading("Lecture Transcription", level=1)
|
| 120 |
+
title.alignment = WD_ALIGN_PARAGRAPH.CENTER
|
| 121 |
+
|
| 122 |
+
doc.add_paragraph("")
|
| 123 |
+
|
| 124 |
+
paragraphs = cleaned_text.split("\n\n")
|
| 125 |
+
|
| 126 |
+
for para in paragraphs:
|
| 127 |
+
p = doc.add_paragraph(para)
|
| 128 |
+
p.paragraph_format.space_after = Pt(12)
|
| 129 |
+
|
| 130 |
+
word_buffer = BytesIO()
|
| 131 |
+
doc.save(word_buffer)
|
| 132 |
+
word_buffer.seek(0)
|
| 133 |
+
return word_buffer
|
| 134 |
+
|
| 135 |
# ---------------------------------------------------
|
| 136 |
+
# FILE UPLOADER
|
| 137 |
# ---------------------------------------------------
|
| 138 |
uploaded_file = st.file_uploader(
|
| 139 |
"Upload Lecture Recording (.mp3, .wav, .m4a, .aac)",
|
| 140 |
+
type=["mp3", "wav", "m4a", "aac"]
|
|
|
|
| 141 |
)
|
| 142 |
|
| 143 |
if uploaded_file:
|
|
|
|
| 153 |
st.info("Loading Whisper model...")
|
| 154 |
model = load_model(model_option)
|
| 155 |
|
|
|
|
| 156 |
start_time = time.time()
|
| 157 |
|
| 158 |
with st.spinner("Transcribing... Please wait."):
|
| 159 |
result = model.transcribe(temp_audio_path)
|
|
|
|
| 160 |
|
| 161 |
end_time = time.time()
|
|
|
|
| 162 |
os.remove(temp_audio_path)
|
| 163 |
|
| 164 |
detected_lang = result.get("language", "Unknown")
|
|
|
|
| 173 |
word_count = len(cleaned_text.split())
|
| 174 |
processing_time = round(end_time - start_time, 2)
|
| 175 |
|
|
|
|
|
|
|
|
|
|
| 176 |
col1, col2 = st.columns(2)
|
| 177 |
|
| 178 |
with col1:
|
|
|
|
| 180 |
st.text_area("", full_text, height=350)
|
| 181 |
|
| 182 |
with col2:
|
| 183 |
+
st.subheader("✨ Clean Story Format")
|
| 184 |
st.text_area("", cleaned_text, height=350)
|
| 185 |
|
| 186 |
st.divider()
|
|
|
|
| 189 |
st.write(f"**Word Count:** {word_count}")
|
| 190 |
st.write(f"**Processing Time:** {processing_time} seconds")
|
| 191 |
|
| 192 |
+
# ----------------------------
|
| 193 |
+
# EXPORT FILES
|
| 194 |
+
# ----------------------------
|
| 195 |
excel_file = create_excel(segments)
|
| 196 |
+
word_file = create_word_document(cleaned_text)
|
| 197 |
+
|
| 198 |
+
colA, colB = st.columns(2)
|
| 199 |
+
|
| 200 |
+
with colA:
|
| 201 |
+
st.download_button(
|
| 202 |
+
label="📥 Download Excel (.xlsx)",
|
| 203 |
+
data=excel_file,
|
| 204 |
+
file_name="RecToText_Transcription.xlsx",
|
| 205 |
+
mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
|
| 206 |
+
)
|
| 207 |
+
|
| 208 |
+
with colB:
|
| 209 |
+
st.download_button(
|
| 210 |
+
label="📄 Download Word (.docx)",
|
| 211 |
+
data=word_file,
|
| 212 |
+
file_name="RecToText_Lecture.docx",
|
| 213 |
+
mime="application/vnd.openxmlformats-officedocument.wordprocessingml.document"
|
| 214 |
+
)
|
| 215 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 216 |
st.divider()
|
| 217 |
st.markdown(
|
| 218 |
+
"<p style='text-align:center; font-size:12px;'>RecToText Pro | Whisper + Streamlit Powered</p>",
|
| 219 |
unsafe_allow_html=True
|
| 220 |
)
|