| |
| import argparse, os, sys, time |
| import numpy as np, soundfile as sf |
| sys.path.insert(0, os.path.dirname(os.path.abspath(__file__))) |
| from audiosr_sdk import AudioSR |
|
|
| def main(): |
| p=argparse.ArgumentParser(description="AudioSR speech super-resolution on AX650") |
| p.add_argument("input") |
| p.add_argument("-o","--output",default=None) |
| p.add_argument("--models",default=os.path.join(os.path.dirname(__file__),"..","models")) |
| p.add_argument("--ddim",default=os.path.join(os.path.dirname(__file__),"..","models","audiosr_ddim_10.npz")) |
| p.add_argument("--seed",type=int,default=42) |
| a=p.parse_args() |
| m=AudioSR(os.path.abspath(a.models),seed=a.seed,ddim=os.path.abspath(a.ddim)) |
| t=time.time(); y,sr=m.enhance(a.input); dt=time.time()-t |
| out=a.output or os.path.splitext(a.input)[0]+"_48k.wav" |
| sf.write(out,y,sr,subtype="PCM_16") |
| d=len(y)/sr |
| print(f"AudioSR: {sr} Hz, {d:.2f}s, {dt:.2f}s inference, RTF {dt/max(d,1e-9):.3f}") |
| print("output:",out) |
|
|
| if __name__=="__main__": main() |
|
|
