app.py

import librosa
import io
import uvicorn
from fastapi import FastAPI, Form, UploadFile, File
from faster_whisper import WhisperModel

# #https://discuss.ray.io/t/expose-deployments/11930
#

### ray ###

app = FastAPI()

model = WhisperModel('tiny', device="cpu", compute_type="int8")

@app.get('/')
def setup():
    return 'hello world'

@app.post('/speech')
async def transcribe(file: UploadFile = File(...)):
    file_bytes = file.file.read()
    audio, sample_rate = librosa.load(io.BytesIO(file_bytes))
    segments, _ = model.transcribe(audio, vad_filter=True)
    return ' '.join([segment.text for segment in segments])

if __name__ == '__main__':
    uvicorn.run('app:app', host='0.0.0.0', port=7860)