egg/Meeting_Assistant
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

OK

Browse Source
This commit is contained in:
beabigegg
2025-11-13 08:18:15 +08:00
parent 788e2409df
commit df5411e44c
38 changed files with 1163 additions and 445 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
tasks.py
View File

@@ -165,21 +165,49 @@ def extract_audio_task(self, input_path, output_path):
@celery.task(base=ProgressTask, bind=True)
def transcribe_audio_task(self, audio_path):
from app import app
import logging
logger = logging.getLogger(__name__)
logger.error(f"[TRANSCRIBE DEBUG] Starting transcribe task for: {audio_path}")
with app.app_context():
try:
logger.error(f"[TRANSCRIBE DEBUG] Entered app context")
self.update_progress(0, 100, "Loading and preparing audio file...")
audio = AudioSegment.from_file(audio_path)
logger.error(f"[TRANSCRIBE DEBUG] Progress updated to 0%")
logger.error(f"[TRANSCRIBE DEBUG] About to load audio file: {audio_path}")
audio = AudioSegment.from_file(audio_path)
logger.error(f"[TRANSCRIBE DEBUG] Audio loaded successfully, duration: {len(audio)}ms")
# 1. Split audio by silence (skip for very long audio to avoid timeout)
logger.error(f"[TRANSCRIBE DEBUG] Starting silence detection")
audio_duration_minutes = len(audio) / (1000 * 60) # Convert to minutes
logger.error(f"[TRANSCRIBE DEBUG] Audio duration: {audio_duration_minutes:.2f} minutes")
if audio_duration_minutes > 10: # Skip silence detection for audio longer than 10 minutes
logger.error(f"[TRANSCRIBE DEBUG] Audio too long ({audio_duration_minutes:.2f} min), skipping silence detection")
self.update_progress(10, 100, f"Audio is {audio_duration_minutes:.1f} minutes long, processing as single chunk...")
chunks = [audio] # Use entire audio as single chunk
else:
self.update_progress(5, 100, "Detecting silence to split audio into chunks...")
try:
chunks = split_on_silence(
audio,
min_silence_len=700,
silence_thresh=-40,
keep_silence=300
)
logger.error(f"[TRANSCRIBE DEBUG] Silence detection completed, found {len(chunks)} chunks")
except Exception as e:
logger.error(f"[TRANSCRIBE DEBUG] Error in silence detection: {str(e)}")
chunks = [audio]
# 1. Split audio by silence
self.update_progress(5, 100, "Detecting silence to split audio into chunks...")
chunks = split_on_silence(
audio,
min_silence_len=700,
silence_thresh=-40,
keep_silence=300
)
if not chunks: # If no silence is detected, treat the whole audio as one chunk
logger.error(f"[TRANSCRIBE DEBUG] No chunks detected, using full audio")
chunks = [audio]
else:
logger.error(f"[TRANSCRIBE DEBUG] Using {len(chunks)} chunks")
# 2. Process chunks and ensure they are within API limits
final_segments = []
@@ -229,10 +257,13 @@ def transcribe_audio_task(self, audio_path):
return {'status': 'Success', 'content': full_content, 'result_path': transcript_filename}
except Exception as e:
import traceback
logger.error(f"[TRANSCRIBE DEBUG] Exception occurred: {type(e).__name__}: {str(e)}")
logger.error(f"[TRANSCRIBE DEBUG] Full traceback: {traceback.format_exc()}")
error_message = f"An error occurred: {str(e)}"
self.update_state(
state='FAILURE',
meta={'exc_type': type(e).__name__, 'exc_message': error_message}
meta={'exc_type': type(e).__name__, 'exc_message': error_message, 'traceback': traceback.format_exc()}
)
return {'status': 'Error', 'error': error_message}