How divyanshu1305/atc-recording-transcription works
This is an ATC-focused speech transcription product: users submit audio through a web console or run a CLI, and receive segmented transcripts with processing metadata and saved artifacts. The documented processing stages are audio cleanup, speech detection and chunking, then ASR; the API also persists results for history queries. Source excerpts establish the API-to-pipeline and API-to-database calls, while the README documents the stage order and capabilities. Some pipeline internals and frontend behavior are only partially sampled.
Main entry points: app.py FastAPI endpoints · transcribe.jsx Transcription console · transcribe_atc.py Pipeline orchestration · run_transcription.py Batch transcription CLI
Components 12
Web console
- Transcription console
frontend/src/components/transcribe.jsx
API and persistence
- FastAPI endpoints
app.py - History queries
app.py - Transcription store
database.py - Audio and JSON artifacts
app.py
Transcription pipeline
- Pipeline orchestration
transcribe_atc.py - Audio cleanup
transcribe_atc.py - VAD and chunking
transcribe_atc.py - ASR and biasing
transcribe_atc.py
Batch processing
- Batch transcription CLI
run_transcription.py
Other components
- Audio operator
- CrispASR executable
Connections 15
- Audio operator → Transcription console: submits audio
README.md - Transcription console → FastAPI endpoints: uploads audio
frontend/src/components/transcribe.jsx - FastAPI endpoints → Pipeline orchestration: processes audio
app.py - Pipeline orchestration → Audio cleanup: preprocesses
README.md - Audio cleanup → VAD and chunking: feeds speech
README.md - VAD and chunking → ASR and biasing: feeds chunks
README.md - ASR and biasing → CrispASR executable: runs inference
transcribe_atc.py - FastAPI endpoints → Transcription store: saves result
app.py - FastAPI endpoints → History queries: serves queries
app.py - History queries → Transcription store: reads records
app.py - FastAPI endpoints → Audio and JSON artifacts: writes artifacts
app.py - Transcription console → History queries: fetches history
frontend/src/components/transcribe.jsx - Transcription console → Audio and JSON artifacts: loads audio
frontend/src/components/transcribe.jsx - Audio operator → Batch transcription CLI: runs batch job
README.md - Batch transcription CLI → Pipeline orchestration: uses pipeline
run_transcription.py