File size: 1,865 Bytes
679fd42 c66ce4c 679fd42 c66ce4c 679fd42 c66ce4c 679fd42 6136d9d 679fd42 |
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 |
from fastapi import FastAPI, UploadFile, File
from fastapi.responses import JSONResponse
from transformers import pipeline
import traceback
import re
import uvicorn
import torchaudio
# ✅ Force torchaudio to use the soundfile backend
torchaudio.set_audio_backend("soundfile")
app = FastAPI(title="Tacab ASR Somali API")
# Load ASR model
asr = pipeline(
"automatic-speech-recognition",
model="tacab/ASR_SOMALI",
tokenizer="tacab/ASR_SOMALI",
chunk_length_s=30,
stride_length_s=6,
return_timestamps="word",
device=-1
)
# Auto punctuation
def auto_punctuate(text):
text = text.strip()
def capitalize_sentences(text):
sentences = re.split(r'(?<=[.?!])\s+', text)
return '. '.join(s.strip().capitalize() for s in sentences if s)
if '.' not in text and len(text.split()) > 5:
text += '.'
words = text.split()
new_text = ""
for i in range(0, len(words), 10):
segment = " ".join(words[i:i+10])
new_text += segment.strip().capitalize() + ". "
return capitalize_sentences(new_text.strip())
@app.post("/transcribe", tags=["ASR"], operation_id="generate")
async def transcribe(file: UploadFile = File(...)):
try:
temp_path = f"/tmp/{file.filename}"
with open(temp_path, "wb") as f:
f.write(await file.read())
result = asr(temp_path)
raw_text = result.get("text", "").strip()
if not raw_text:
return JSONResponse({"error": "No transcription result."}, status_code=400)
cleaned_text = auto_punctuate(raw_text)
# return {"transcription": cleaned_text}
return {"text": cleaned_text}
except Exception as e:
traceback.print_exc()
return JSONResponse({"error": str(e)}, status_code=500)
if __name__ == "__main__":
uvicorn.run(app, host="0.0.0.0", port=7860)
|