Spaces:
Running
Running
nice - downloads too
Browse files- .gitignore +2 -1
- app.py +52 -18
.gitignore
CHANGED
@@ -1,3 +1,4 @@
|
|
1 |
.venv/
|
2 |
transcripts/
|
3 |
-
temp_audio/
|
|
|
|
1 |
.venv/
|
2 |
transcripts/
|
3 |
+
temp_audio/
|
4 |
+
temp_downloads/
|
app.py
CHANGED
@@ -11,6 +11,7 @@ from pydub import AudioSegment
|
|
11 |
import asyncio
|
12 |
import io
|
13 |
from itertools import groupby
|
|
|
14 |
|
15 |
prompt = '''
|
16 |
You are an expert transcript editor. Your task is to enhance this transcript for maximum readability while maintaining the core message.
|
@@ -256,6 +257,23 @@ def rename_speakers(text: str, speaker_map: dict) -> str:
|
|
256 |
return result
|
257 |
|
258 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
259 |
def process_audio(audio_file):
|
260 |
try:
|
261 |
temp_path = Path("temp_audio")
|
@@ -266,11 +284,12 @@ def process_audio(audio_file):
|
|
266 |
with open(temp_file, "wb") as f:
|
267 |
f.write(audio_file)
|
268 |
|
269 |
-
# Initial state -
|
270 |
yield (
|
271 |
gr.update(value="", visible=True), # original transcript
|
272 |
gr.update(value="", visible=True), # enhanced transcript
|
273 |
-
|
|
|
274 |
)
|
275 |
|
276 |
# Get transcript
|
@@ -279,11 +298,15 @@ def process_audio(audio_file):
|
|
279 |
dialogues = list(group_utterances_by_speaker(utterances))
|
280 |
original = format_chunk(dialogues, markdown=True)
|
281 |
|
282 |
-
#
|
|
|
|
|
|
|
283 |
yield (
|
284 |
gr.update(value=original, visible=True),
|
285 |
gr.update(value="", visible=True),
|
286 |
-
|
|
|
287 |
)
|
288 |
|
289 |
try:
|
@@ -293,21 +316,27 @@ def process_audio(audio_file):
|
|
293 |
merged = "\n\n".join(chunk.strip() for chunk in enhanced)
|
294 |
merged = apply_markdown_formatting(merged)
|
295 |
|
|
|
|
|
|
|
296 |
# Show final result
|
297 |
yield (
|
298 |
gr.update(value=original, visible=True),
|
299 |
gr.update(value=merged, visible=True),
|
300 |
-
|
|
|
301 |
)
|
302 |
|
303 |
except Exception as e:
|
304 |
yield (
|
305 |
gr.update(value=original, visible=True),
|
306 |
gr.update(value=f"Error: {str(e)}", visible=True),
|
307 |
-
|
|
|
308 |
)
|
309 |
|
310 |
finally:
|
|
|
311 |
if os.path.exists(temp_file):
|
312 |
os.remove(temp_file)
|
313 |
|
@@ -342,36 +371,41 @@ with gr.Blocks(title="Transcript Enhancer") as demo:
|
|
342 |
with gr.Row():
|
343 |
with gr.Column():
|
344 |
gr.Markdown("### Original Transcript")
|
|
|
|
|
|
|
|
|
|
|
|
|
345 |
original_output = gr.Markdown()
|
346 |
|
347 |
with gr.Column():
|
348 |
gr.Markdown("### Enhanced Transcript")
|
349 |
-
|
350 |
-
|
351 |
-
|
|
|
|
|
352 |
)
|
353 |
enhanced_output = gr.Markdown()
|
354 |
|
355 |
-
# Add some CSS
|
356 |
gr.Markdown("""
|
357 |
<style>
|
358 |
-
.
|
359 |
-
|
360 |
-
border-radius: 4px;
|
361 |
-
background-color: #f0f0f0;
|
362 |
-
margin-bottom: 10px;
|
363 |
-
display: inline-block;
|
364 |
}
|
365 |
</style>
|
366 |
""")
|
367 |
-
|
368 |
transcribe_btn.click(
|
369 |
fn=process_audio,
|
370 |
inputs=[audio_input],
|
371 |
outputs=[
|
372 |
original_output,
|
373 |
enhanced_output,
|
374 |
-
|
|
|
375 |
]
|
376 |
)
|
377 |
|
|
|
11 |
import asyncio
|
12 |
import io
|
13 |
from itertools import groupby
|
14 |
+
from datetime import datetime
|
15 |
|
16 |
prompt = '''
|
17 |
You are an expert transcript editor. Your task is to enhance this transcript for maximum readability while maintaining the core message.
|
|
|
257 |
return result
|
258 |
|
259 |
|
260 |
+
def create_downloadable_file(content: str, prefix: str) -> str:
|
261 |
+
"""Create a temporary file with the content and return filepath"""
|
262 |
+
temp_dir = Path("temp_downloads")
|
263 |
+
temp_dir.mkdir(exist_ok=True)
|
264 |
+
|
265 |
+
# Create a unique filename
|
266 |
+
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
267 |
+
filename = f"{prefix}_{timestamp}.md"
|
268 |
+
filepath = temp_dir / filename
|
269 |
+
|
270 |
+
# Write content to file
|
271 |
+
with open(filepath, "w", encoding="utf-8") as f:
|
272 |
+
f.write(content)
|
273 |
+
|
274 |
+
return str(filepath)
|
275 |
+
|
276 |
+
|
277 |
def process_audio(audio_file):
|
278 |
try:
|
279 |
temp_path = Path("temp_audio")
|
|
|
284 |
with open(temp_file, "wb") as f:
|
285 |
f.write(audio_file)
|
286 |
|
287 |
+
# Initial state - clear both transcripts
|
288 |
yield (
|
289 |
gr.update(value="", visible=True), # original transcript
|
290 |
gr.update(value="", visible=True), # enhanced transcript
|
291 |
+
None, # original download
|
292 |
+
None, # enhanced download
|
293 |
)
|
294 |
|
295 |
# Get transcript
|
|
|
298 |
dialogues = list(group_utterances_by_speaker(utterances))
|
299 |
original = format_chunk(dialogues, markdown=True)
|
300 |
|
301 |
+
# Create downloadable file for original transcript
|
302 |
+
original_file = create_downloadable_file(original, "original_transcript")
|
303 |
+
|
304 |
+
# Show original transcript
|
305 |
yield (
|
306 |
gr.update(value=original, visible=True),
|
307 |
gr.update(value="", visible=True),
|
308 |
+
original_file,
|
309 |
+
None,
|
310 |
)
|
311 |
|
312 |
try:
|
|
|
316 |
merged = "\n\n".join(chunk.strip() for chunk in enhanced)
|
317 |
merged = apply_markdown_formatting(merged)
|
318 |
|
319 |
+
# Create downloadable file for enhanced transcript
|
320 |
+
enhanced_file = create_downloadable_file(merged, "enhanced_transcript")
|
321 |
+
|
322 |
# Show final result
|
323 |
yield (
|
324 |
gr.update(value=original, visible=True),
|
325 |
gr.update(value=merged, visible=True),
|
326 |
+
original_file,
|
327 |
+
enhanced_file,
|
328 |
)
|
329 |
|
330 |
except Exception as e:
|
331 |
yield (
|
332 |
gr.update(value=original, visible=True),
|
333 |
gr.update(value=f"Error: {str(e)}", visible=True),
|
334 |
+
original_file,
|
335 |
+
None,
|
336 |
)
|
337 |
|
338 |
finally:
|
339 |
+
# Cleanup temp files
|
340 |
if os.path.exists(temp_file):
|
341 |
os.remove(temp_file)
|
342 |
|
|
|
371 |
with gr.Row():
|
372 |
with gr.Column():
|
373 |
gr.Markdown("### Original Transcript")
|
374 |
+
original_download = gr.File(
|
375 |
+
label="Download as Markdown",
|
376 |
+
file_count="single",
|
377 |
+
visible=True,
|
378 |
+
interactive=False,
|
379 |
+
)
|
380 |
original_output = gr.Markdown()
|
381 |
|
382 |
with gr.Column():
|
383 |
gr.Markdown("### Enhanced Transcript")
|
384 |
+
enhanced_download = gr.File(
|
385 |
+
label="Download as Markdown",
|
386 |
+
file_count="single",
|
387 |
+
visible=True,
|
388 |
+
interactive=False,
|
389 |
)
|
390 |
enhanced_output = gr.Markdown()
|
391 |
|
392 |
+
# Add some CSS to style the download buttons
|
393 |
gr.Markdown("""
|
394 |
<style>
|
395 |
+
.download-button {
|
396 |
+
margin-top: 10px;
|
|
|
|
|
|
|
|
|
397 |
}
|
398 |
</style>
|
399 |
""")
|
400 |
+
|
401 |
transcribe_btn.click(
|
402 |
fn=process_audio,
|
403 |
inputs=[audio_input],
|
404 |
outputs=[
|
405 |
original_output,
|
406 |
enhanced_output,
|
407 |
+
original_download,
|
408 |
+
enhanced_download
|
409 |
]
|
410 |
)
|
411 |
|