File size: 5,488 Bytes
e275025 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 |
#!/usr/bin/env python3
"""
Trans-for-Doctors CLI
Runs the end-to-end pipeline: STT → Knowledge Base → LLM Correction → (optional) DOCX report.
Usage examples:
uv run transmed --audio path/to.wav --model . --llm --generate-report
uv run transmed --audio path/to.wav --model . --no-llm
"""
import argparse
import logging
import os
from pathlib import Path
from pipeline import MedicalTranscriptionPipeline, PipelineConfig
def setup_logging(level: str = "INFO") -> None:
logging.basicConfig(
level=getattr(logging, level.upper(), logging.INFO),
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(
description="Run medical transcription pipeline (STT + LLM Corrector + KB)",
)
# Core
parser.add_argument("--audio", required=True, type=str, help="Path to audio .wav file")
parser.add_argument("--model", type=str, default=".", help="Path to Whisper model directory")
parser.add_argument("--device", type=str, default="auto", choices=["auto", "cuda", "cpu", "mps"], help="Inference device")
parser.add_argument("--dtype", type=str, default="float32", choices=["float32", "float16", "bfloat16"], help="Torch dtype")
parser.add_argument("--language", type=str, default="russian", help="Transcription language")
# Knowledge base
parser.add_argument("--terms", type=str, default="medical_terms.txt", help="Path to medical terms file")
# LLM correction
parser.add_argument("--llm", dest="llm", action="store_true", help="Enable LLM correction")
parser.add_argument("--no-llm", dest="llm", action="store_false", help="Disable LLM correction")
parser.set_defaults(llm=True)
parser.add_argument("--openai-model", type=str, default="gpt-4o", help="OpenAI model name")
parser.add_argument("--openai-key", type=str, default=os.getenv("OPENAI_API_KEY"), help="OpenAI API key (defaults to env OPENAI_API_KEY)")
# Outputs
parser.add_argument("--save-original", action="store_true", help="Save original transcription JSON")
parser.add_argument("--save-corrected", action="store_true", help="Save corrected transcription JSON")
parser.add_argument("--generate-report", action="store_true", help="Generate DOCX report")
parser.add_argument("--results-dir", type=str, default="results", help="Directory to store results")
parser.add_argument("--logs-dir", type=str, default="logs", help="Directory to store logs")
# Logging
parser.add_argument("--log-level", type=str, default="INFO", help="Logging level")
# Patient metadata (optional)
parser.add_argument("--patient-name", type=str, default=None)
parser.add_argument("--patient-id", type=str, default=None)
parser.add_argument("--study-date", type=str, default=None)
parser.add_argument("--modality", type=str, default=None)
parser.add_argument("--body-part", type=str, default=None)
return parser.parse_args()
def main() -> None:
args = parse_args()
setup_logging(args.log_level)
logger = logging.getLogger("transmed")
audio_path = Path(args.audio)
model_path = Path(args.model)
terms_path = Path(args.terms)
results_dir = Path(args.results_dir)
logs_dir = Path(args.logs_dir)
if not audio_path.exists():
logger.error(f"Audio file not found: {audio_path}")
raise SystemExit(1)
if not model_path.exists():
logger.error(f"Model path not found: {model_path}")
raise SystemExit(1)
if not terms_path.exists():
logger.warning(f"Terms file not found: {terms_path} — proceeding without extra terms")
# Configure pipeline
config = PipelineConfig(
model_path=model_path,
device=args.device,
dtype=args.dtype,
language=args.language,
medical_terms_file=terms_path,
openai_api_key=args.openai_key,
openai_model=args.openai_model,
correction_enabled=args.llm,
save_original=args.save_original,
save_corrected=args.save_corrected,
save_diff=True,
generate_report=args.generate_report,
results_dir=results_dir,
reports_dir=results_dir / "reports",
logs_dir=logs_dir,
)
logger.info("Creating medical transcription pipeline...")
pipeline = MedicalTranscriptionPipeline(config)
patient_metadata = None
if args.generate_report:
patient_metadata = {
"patient_name": args.patient_name,
"patient_id": args.patient_id,
"study_date": args.study_date,
"modality": args.modality,
"body_part": args.body_part,
}
logger.info(f"Processing audio: {audio_path.name}")
result = pipeline.process_audio_file(audio_path=audio_path, patient_metadata=patient_metadata)
if result.get("status") != "success":
logger.error(f"Pipeline failed: {result.get('error')}")
raise SystemExit(2)
# Summarize
orig = result.get("original_transcription", "")
corr = result.get("corrected_transcription", orig)
logger.info(f"Original ({len(orig)} chars): {orig[:200]}...")
if config.correction_enabled:
logger.info(f"Corrected ({len(corr)} chars): {corr[:200]}...")
logger.info(f"Corrections: {len(result.get('corrections', []))}")
if result.get("report_path"):
logger.info(f"Report: {result['report_path']}")
if __name__ == "__main__":
main()
|