Vishal Patil commited on
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,9 +1,5 @@
|
|
| 1 |
-
"""
|
| 2 |
-
Gradio Web UI for Speech Annotation Pipeline
|
| 3 |
-
"""
|
| 4 |
-
|
| 5 |
-
import json
|
| 6 |
import tempfile
|
|
|
|
| 7 |
from pathlib import Path
|
| 8 |
import gradio as gr
|
| 9 |
from pipeline import get_pipeline, to_json, to_csv, to_abab_text
|
|
@@ -15,21 +11,16 @@ _last_segments = []
|
|
| 15 |
|
| 16 |
def process_audio(audio_path):
|
| 17 |
global _last_segments
|
| 18 |
-
|
| 19 |
if audio_path is None:
|
| 20 |
-
return "⚠️ Please upload an audio file first.", "", ""
|
| 21 |
-
|
| 22 |
try:
|
| 23 |
pipeline = get_pipeline()
|
| 24 |
segments = pipeline.process(audio_path)
|
| 25 |
-
except EnvironmentError as e:
|
| 26 |
-
return f"❌ Config error: {e}", "", "", None
|
| 27 |
except Exception as e:
|
| 28 |
-
|
| 29 |
-
return f"❌ Error: {e}\n{traceback.format_exc()}", "", "", None
|
| 30 |
|
| 31 |
if not segments:
|
| 32 |
-
return "⚠️ No speech detected.", "", ""
|
| 33 |
|
| 34 |
_last_segments = segments
|
| 35 |
abab = to_abab_text(segments)
|
|
@@ -37,84 +28,58 @@ def process_audio(audio_path):
|
|
| 37 |
rows = "".join(
|
| 38 |
f"<tr style='background:{'#f0f4ff' if i%2==0 else 'white'}'>"
|
| 39 |
f"<td style='padding:6px 8px;font-weight:bold;color:#1e3a5f'>{s.speaker}</td>"
|
| 40 |
-
f"<td style='padding:6px 8px
|
| 41 |
-
f"<td style='padding:6px 8px
|
| 42 |
f"<td style='padding:6px 8px'>{s.text}</td></tr>"
|
| 43 |
for i, s in enumerate(segments)
|
| 44 |
)
|
| 45 |
-
table = f"""
|
| 46 |
-
<table style='width:100%;border-collapse:collapse;font-family:monospace;font-size:14px'>
|
| 47 |
<thead style='background:#1e3a5f;color:white'>
|
| 48 |
-
<tr>
|
| 49 |
-
|
| 50 |
-
<th style='padding:8px'>Start</th>
|
| 51 |
-
<th style='padding:8px'>End</th>
|
| 52 |
-
<th style='padding:8px;text-align:left'>Transcript</th>
|
| 53 |
-
</tr>
|
| 54 |
-
</thead>
|
| 55 |
-
<tbody>{rows}</tbody>
|
| 56 |
-
</table>
|
| 57 |
-
"""
|
| 58 |
|
| 59 |
-
|
| 60 |
-
return status, abab, table, None
|
| 61 |
|
| 62 |
|
| 63 |
def export_json():
|
| 64 |
if not _last_segments:
|
| 65 |
return None
|
| 66 |
-
out = str(UPLOAD_DIR / "
|
| 67 |
to_json(_last_segments, out)
|
| 68 |
return out
|
| 69 |
|
| 70 |
-
|
| 71 |
def export_csv():
|
| 72 |
if not _last_segments:
|
| 73 |
return None
|
| 74 |
-
out = str(UPLOAD_DIR / "
|
| 75 |
to_csv(_last_segments, out)
|
| 76 |
return out
|
| 77 |
|
| 78 |
|
| 79 |
with gr.Blocks(title="Speech Annotation Pipeline") as demo:
|
| 80 |
-
|
| 81 |
-
gr.
|
| 82 |
-
gr.HTML("<p style='text-align:center;color:#666'>Upload audio → get speaker-labeled transcripts automatically</p>")
|
| 83 |
|
| 84 |
with gr.Row():
|
| 85 |
-
with gr.Column(
|
| 86 |
-
audio_input = gr.Audio(
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
)
|
| 90 |
-
run_btn = gr.Button("▶ Run Annotation", variant="primary", size="lg")
|
| 91 |
-
status_box = gr.Textbox(label="Status", interactive=False, lines=2)
|
| 92 |
-
|
| 93 |
-
gr.Markdown("### Export Results")
|
| 94 |
with gr.Row():
|
| 95 |
json_btn = gr.Button("Export JSON")
|
| 96 |
csv_btn = gr.Button("Export CSV")
|
| 97 |
-
json_file = gr.File(label="JSON
|
| 98 |
-
csv_file = gr.File(label="CSV
|
| 99 |
-
|
| 100 |
-
with gr.Column(
|
| 101 |
-
gr.
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
gr.Markdown("### Segment Table")
|
| 108 |
-
table_html = gr.HTML()
|
| 109 |
-
|
| 110 |
-
# Wire up button clicks
|
| 111 |
-
run_btn.click(
|
| 112 |
-
fn=process_audio,
|
| 113 |
-
inputs=[audio_input],
|
| 114 |
-
outputs=[status_box, abab_box, table_html, json_file],
|
| 115 |
-
)
|
| 116 |
json_btn.click(fn=export_json, inputs=[], outputs=[json_file])
|
| 117 |
csv_btn.click(fn=export_csv, inputs=[], outputs=[csv_file])
|
| 118 |
|
| 119 |
if __name__ == "__main__":
|
| 120 |
-
demo.launch(server_name="0.0.0.0", server_port=7860
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import tempfile
|
| 2 |
+
import traceback
|
| 3 |
from pathlib import Path
|
| 4 |
import gradio as gr
|
| 5 |
from pipeline import get_pipeline, to_json, to_csv, to_abab_text
|
|
|
|
| 11 |
|
| 12 |
def process_audio(audio_path):
|
| 13 |
global _last_segments
|
|
|
|
| 14 |
if audio_path is None:
|
| 15 |
+
return "⚠️ Please upload an audio file first.", "", ""
|
|
|
|
| 16 |
try:
|
| 17 |
pipeline = get_pipeline()
|
| 18 |
segments = pipeline.process(audio_path)
|
|
|
|
|
|
|
| 19 |
except Exception as e:
|
| 20 |
+
return f"❌ Error: {e}\n{traceback.format_exc()}", "", ""
|
|
|
|
| 21 |
|
| 22 |
if not segments:
|
| 23 |
+
return "⚠️ No speech detected.", "", ""
|
| 24 |
|
| 25 |
_last_segments = segments
|
| 26 |
abab = to_abab_text(segments)
|
|
|
|
| 28 |
rows = "".join(
|
| 29 |
f"<tr style='background:{'#f0f4ff' if i%2==0 else 'white'}'>"
|
| 30 |
f"<td style='padding:6px 8px;font-weight:bold;color:#1e3a5f'>{s.speaker}</td>"
|
| 31 |
+
f"<td style='padding:6px 8px'>{s.start_fmt}</td>"
|
| 32 |
+
f"<td style='padding:6px 8px'>{s.end_fmt}</td>"
|
| 33 |
f"<td style='padding:6px 8px'>{s.text}</td></tr>"
|
| 34 |
for i, s in enumerate(segments)
|
| 35 |
)
|
| 36 |
+
table = f"""<table style='width:100%;border-collapse:collapse;font-size:14px'>
|
|
|
|
| 37 |
<thead style='background:#1e3a5f;color:white'>
|
| 38 |
+
<tr><th style='padding:8px'>Speaker</th><th>Start</th><th>End</th><th style='text-align:left;padding:8px'>Transcript</th></tr>
|
| 39 |
+
</thead><tbody>{rows}</tbody></table>"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
+
return f"✅ Done — {len(segments)} segments", abab, table
|
|
|
|
| 42 |
|
| 43 |
|
| 44 |
def export_json():
|
| 45 |
if not _last_segments:
|
| 46 |
return None
|
| 47 |
+
out = str(UPLOAD_DIR / "annotation.json")
|
| 48 |
to_json(_last_segments, out)
|
| 49 |
return out
|
| 50 |
|
|
|
|
| 51 |
def export_csv():
|
| 52 |
if not _last_segments:
|
| 53 |
return None
|
| 54 |
+
out = str(UPLOAD_DIR / "annotation.csv")
|
| 55 |
to_csv(_last_segments, out)
|
| 56 |
return out
|
| 57 |
|
| 58 |
|
| 59 |
with gr.Blocks(title="Speech Annotation Pipeline") as demo:
|
| 60 |
+
gr.Markdown("# 🎙️ Speech Annotation Pipeline")
|
| 61 |
+
gr.Markdown("Upload audio → get speaker-labeled transcripts automatically")
|
|
|
|
| 62 |
|
| 63 |
with gr.Row():
|
| 64 |
+
with gr.Column():
|
| 65 |
+
audio_input = gr.Audio(label="Upload Audio (.wav/.mp3/.flac)", type="filepath")
|
| 66 |
+
run_btn = gr.Button("▶ Run Annotation", variant="primary")
|
| 67 |
+
status_box = gr.Textbox(label="Status", interactive=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
with gr.Row():
|
| 69 |
json_btn = gr.Button("Export JSON")
|
| 70 |
csv_btn = gr.Button("Export CSV")
|
| 71 |
+
json_file = gr.File(label="JSON")
|
| 72 |
+
csv_file = gr.File(label="CSV")
|
| 73 |
+
|
| 74 |
+
with gr.Column():
|
| 75 |
+
abab_box = gr.Textbox(label="Editable Transcript (ABAB)", lines=20,
|
| 76 |
+
placeholder="Annotations will appear here...")
|
| 77 |
+
table_html = gr.HTML(label="Segment Table")
|
| 78 |
+
|
| 79 |
+
run_btn.click(fn=process_audio, inputs=[audio_input],
|
| 80 |
+
outputs=[status_box, abab_box, table_html])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
json_btn.click(fn=export_json, inputs=[], outputs=[json_file])
|
| 82 |
csv_btn.click(fn=export_csv, inputs=[], outputs=[csv_file])
|
| 83 |
|
| 84 |
if __name__ == "__main__":
|
| 85 |
+
demo.launch(server_name="0.0.0.0", server_port=7860)
|