Skip to content

Speakr is a personal, self-hosted web application designed for transcribing audio recordings

Python

Fix audio chunking not working after refactor (#140)

Murtaza Nasir6afc2dbfc4b0
diff --git a/src/config/app_config.py b/src/config/app_config.py
index 8938aaeb..dcc75a37 100644
--- a/src/config/app_config.py
+++ b/src/config/app_config.py
@@ -40,6 +40,9 @@ ENABLE_CHUNKING = os.environ.get('ENABLE_CHUNKING', 'true').lower() == 'true'
CHUNK_SIZE_MB = int(os.environ.get('CHUNK_SIZE_MB', '20'))
CHUNK_OVERLAP_SECONDS = int(os.environ.get('CHUNK_OVERLAP_SECONDS', '3'))
+# Create chunking service at module level so it can be imported by processing.py
+chunking_service = AudioChunkingService(CHUNK_SIZE_MB, CHUNK_OVERLAP_SECONDS) if ENABLE_CHUNKING else None
+
def initialize_config(app):
"""Initialize application configuration."""
@@ -59,7 +62,7 @@ def initialize_config(app):
except Exception as e:
app.logger.error(f"Failed to initialize LLM client: {e}")
- chunking_service = AudioChunkingService(CHUNK_SIZE_MB, CHUNK_OVERLAP_SECONDS) if ENABLE_CHUNKING else None
+ # Use module-level chunking_service (already created above)
version = get_version()
app.logger.info(f"=== Speakr {version} Starting Up ===")
diff --git a/src/tasks/processing.py b/src/tasks/processing.py
index c96c7533..7481350a 100644
--- a/src/tasks/processing.py
+++ b/src/tasks/processing.py
@@ -18,6 +18,7 @@ import subprocess
import httpx
from datetime import datetime
from flask import current_app
+from openai import OpenAI
from src.database import db
from src.models import Recording, Tag, Event, TranscriptChunk, SystemSetting, GroupMembership, RecordingTag, InternalShare, SharedRecordingState, User
@@ -25,7 +26,7 @@ from src.services.embeddings import process_recording_chunks
from src.services.llm import is_using_openai_api, call_llm_completion, format_api_error_message, TEXT_MODEL_NAME, client, http_client_no_proxy
from src.utils import extract_json_object, safe_json_loads
from src.audio_chunking import AudioChunkingService, ChunkProcessingError, ChunkingNotSupportedError
-from src.config.app_config import ASR_DIARIZE, ASR_BASE_URL, transcription_api_key, transcription_base_url
+from src.config.app_config import ASR_DIARIZE, ASR_BASE_URL, transcription_api_key, transcription_base_url, chunking_service, ENABLE_CHUNKING
from src.file_exporter import export_recording, ENABLE_AUTO_EXPORT
# Configuration for internal sharing
@@ -174,20 +175,13 @@ def clean_llm_response(text):
USE_ASR_ENDPOINT = os.environ.get('USE_ASR_ENDPOINT', 'false').lower() == 'true'
ASR_ENDPOINT = os.environ.get('ASR_ENDPOINT', '')
ASR_API_KEY = os.environ.get('ASR_API_KEY', '')
-OPENAI_API_KEY = os.environ.get('OPENAI_API_KEY')
ENABLE_INQUIRE_MODE = os.environ.get('ENABLE_INQUIRE_MODE', 'false').lower() == 'true'
-ENABLE_CHUNKING = os.environ.get('ENABLE_CHUNKING', 'true').lower() == 'true'
-
-# Chunking service will be passed in from app
-chunking_service = None
-# Import OpenAI client
-from openai import OpenAI
+# chunking_service, ENABLE_CHUNKING, transcription_api_key, and transcription_base_url
+# are imported from src.config.app_config
-if USE_ASR_ENDPOINT:
- asr_client = OpenAI(base_url=ASR_ENDPOINT, api_key=ASR_API_KEY)
-else:
- asr_client = OpenAI(api_key=OPENAI_API_KEY)
+# Note: OpenAI clients are created inside each transcription function as needed,
+# not at module level (matching original pre-refactor behavior)
def generate_title_task(app_context, recording_id, will_auto_summarize=False):