import sys, re, numpy as np, subprocess
from faster_whisper import WhisperModel
S = ["大家好,我是方舟老師,歡迎來到這門課。", "機器人其實就是三件事,感測、思考、行動。", "通電之前,一定要先確認電池的接線。"]
norm = lambda s: re.sub(r'[^\w]', '', s)
def lev(a, b):
    d = list(range(len(b)+1))
    for i, ca in enumerate(a, 1):
        p, d[0] = d[0], i
        for j, cb in enumerate(b, 1):
            p, d[j] = d[j], min(d[j]+1, d[j-1]+1, p + (ca != cb))
    return d[-1]
m = WhisperModel("small", device="cpu", compute_type="int8", cpu_threads=6)
for tag in sys.argv[1:]:
    tot = err = 0
    for i, t in enumerate(S):
        pcm = subprocess.run(["ffmpeg","-v","error","-i",f"out/r{tag}-{i}.wav","-ar","16000","-ac","1","-f","f32le","-"],capture_output=True).stdout
        segs,_ = m.transcribe(np.frombuffer(pcm,dtype=np.float32), language="zh", beam_size=5, initial_prompt="以下是繁體中文的課程講解。")
        h = norm("".join(s.text for s in segs)); r = norm(t)
        tot += len(r); err += lev(r, h); print(f"  {tag}-{i}: {h}")
    print(f"{tag} CER {err/tot:.0%}")
