Skip to content

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

Python

Highlight cloud diarization feature and fix ASR timeout fallback

Murtaza Nasird45f48e9cd6a
diff --git a/README.md b/README.md
index ff04f0fc..ec3f4d15 100644
--- a/README.md
+++ b/README.md
@@ -158,14 +158,12 @@ Complete documentation is available at **[murtaza-nasir.github.io/speakr](https:
## Latest Release (v0.8.0)
-**Connector Architecture & REST API**
+**Cloud Diarization & REST API**
-- **Connector-Based Transcription** - New modular architecture for transcription providers; custom connectors possible (adding new providers requires building from source)
-- **Simplified Configuration** - Fewer environment variables needed; auto-detects connector from settings
-- **OpenAI Diarization Support** - Use `gpt-4o-transcribe-diarize` for speaker identification without self-hosting
+- **Speaker Diarization Without GPU** - Use OpenAI's `gpt-4o-transcribe-diarize` for speaker identification with just an API key. No self-hosted ASR container or GPU required.
- **REST API v1** - Full-featured API for automation tools (n8n, Zapier, Make) and dashboard widgets
- **Interactive API Docs** - Swagger UI at `/api/v1/docs` with try-it-out functionality
-- **Batch Operations** - Update, delete, or transcribe multiple recordings via API
+- **Connector Architecture** - Modular transcription providers with simplified configuration; custom connectors possible (requires building from source)
> **Migration Note:** Existing configurations are fully backwards compatible. See the [Migration Guide](https://murtaza-nasir.github.io/speakr/admin-guide/migration-guide) for recommended updates.
diff --git a/src/services/transcription/registry.py b/src/services/transcription/registry.py
index a314862f..9eec4ff4 100644
--- a/src/services/transcription/registry.py
+++ b/src/services/transcription/registry.py
@@ -171,6 +171,34 @@ class ConnectorRegistry:
logger.error(f"Failed to initialize connector '{connector_name}': {e}")
raise ConfigurationError(f"Failed to initialize connector '{connector_name}': {e}") from e
+ def _get_asr_timeout(self) -> int:
+ """
+ Get ASR timeout with fallback chain: ENV -> Admin UI -> default.
+
+ Priority:
+ 1. ASR_TIMEOUT environment variable
+ 2. asr_timeout_seconds environment variable (legacy)
+ 3. SystemSetting from Admin UI (database)
+ 4. Default: 1800 seconds (30 minutes)
+ """
+ # Check environment variables first
+ env_timeout = os.environ.get('ASR_TIMEOUT') or os.environ.get('asr_timeout_seconds')
+ if env_timeout:
+ return int(env_timeout)
+
+ # Fall back to Admin UI setting (SystemSetting in database)
+ try:
+ from src.models import SystemSetting
+ db_timeout = SystemSetting.get_setting('asr_timeout_seconds', None)
+ if db_timeout is not None:
+ return int(db_timeout)
+ except Exception as e:
+ # May fail if no app context or during initialization
+ logger.debug(f"Could not read ASR timeout from database: {e}")
+
+ # Default: 30 minutes
+ return 1800
+
def _build_config_from_env(self, connector_name: str) -> Dict[str, Any]:
"""
Build connector config from environment variables.
@@ -188,7 +216,7 @@ class ConnectorRegistry:
return {
'base_url': base_url,
- 'timeout': int(os.environ.get('ASR_TIMEOUT', os.environ.get('asr_timeout_seconds', '1800'))),
+ 'timeout': self._get_asr_timeout(),
'diarize': os.environ.get('ASR_DIARIZE', 'true').lower() == 'true',
'return_speaker_embeddings': os.environ.get('ASR_RETURN_SPEAKER_EMBEDDINGS', 'false').lower() == 'true'
}