Spaces:
Running
Running
Create transcription_routes.py
Browse files- transcription_routes.py +60 -0
transcription_routes.py
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# transcription_routes.py
|
2 |
+
|
3 |
+
import os
|
4 |
+
import io
|
5 |
+
from fastapi import APIRouter, File, UploadFile, HTTPException, Request
|
6 |
+
from fastapi.responses import JSONResponse
|
7 |
+
from groq import Groq
|
8 |
+
from groq.errors import GroqError
|
9 |
+
|
10 |
+
router = APIRouter()
|
11 |
+
|
12 |
+
# Load your Groq API key
|
13 |
+
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
|
14 |
+
if not GROQ_API_KEY:
|
15 |
+
raise ValueError("GROQ_API_KEY environment variable is not set")
|
16 |
+
|
17 |
+
client = Groq(api_key=GROQ_API_KEY)
|
18 |
+
|
19 |
+
@router.exception_handler(Exception)
|
20 |
+
async def global_exception_handler(request: Request, exc: Exception):
|
21 |
+
return JSONResponse(
|
22 |
+
status_code=500, content={"detail": str(exc)}
|
23 |
+
)
|
24 |
+
|
25 |
+
@router.post(
|
26 |
+
"/transcribe",
|
27 |
+
summary="Upload an audio file and return its transcription",
|
28 |
+
response_description="Returns transcribed text as JSON"
|
29 |
+
)
|
30 |
+
async def transcribe_audio(file: UploadFile = File(...)):
|
31 |
+
if not file.filename:
|
32 |
+
raise HTTPException(status_code=400, detail="No file provided")
|
33 |
+
|
34 |
+
# Only allow common audio formats
|
35 |
+
allowed_exts = (".mp3", ".wav", ".m4a", ".flac")
|
36 |
+
if not file.filename.lower().endswith(allowed_exts):
|
37 |
+
raise HTTPException(
|
38 |
+
status_code=415,
|
39 |
+
detail=f"Unsupported file type. Allowed: {', '.join(allowed_exts)}"
|
40 |
+
)
|
41 |
+
|
42 |
+
data = await file.read()
|
43 |
+
try:
|
44 |
+
# send to Groq Whisper model
|
45 |
+
resp = client.audio.transcriptions.create(
|
46 |
+
file=(file.filename, data),
|
47 |
+
model="whisper-large-v3",
|
48 |
+
response_format="verbose_json",
|
49 |
+
)
|
50 |
+
# the library returns an object with a `.text` attribute
|
51 |
+
transcript = getattr(resp, "text", None)
|
52 |
+
if transcript is None:
|
53 |
+
# fallback if resp is dict-like
|
54 |
+
transcript = resp.get("text") # type: ignore
|
55 |
+
except GroqError as e:
|
56 |
+
raise HTTPException(status_code=502, detail=f"Transcription service error: {e}")
|
57 |
+
except Exception as e:
|
58 |
+
raise HTTPException(status_code=500, detail=f"Unexpected error: {e}")
|
59 |
+
|
60 |
+
return JSONResponse(content={"transcript": transcript})
|