Function compute_wer

funasr/metrics/wer.py:8–121

Compute wer. Args: ref_file: TODO. hyp_file: TODO. cer_file: TODO. cn_postprocess: TODO.

(
    ref_file,
    hyp_file,
    cer_file,
    cn_postprocess=False,
)

def compute_wer(
    ref_file,
    hyp_file,
    cer_file,
    cn_postprocess=False,
):
    """Compute wer.
    
        Args:
            ref_file: TODO.
            hyp_file: TODO.
            cer_file: TODO.
            cn_postprocess: TODO.
        """
    rst = {
        "Wrd": 0,
        "Corr": 0,
        "Ins": 0,
        "Del": 0,
        "Sub": 0,
        "Snt": 0,
        "Err": 0.0,
        "S.Err": 0.0,
        "wrong_words": 0,
        "wrong_sentences": 0,
    }

    hyp_dict = {}
    ref_dict = {}
    with open(hyp_file, "r") as hyp_reader:
        for line in hyp_reader:
            key = line.strip().split()[0]
            value = line.strip().split()[1:]
            if cn_postprocess:
                value = " ".join(value)
                value = value.replace(" ", "")
                # if value[0] == "请":
                #     value = value[1:]
                value = [x for x in value]
            hyp_dict[key] = value
    with open(ref_file, "r") as ref_reader:
        for line in ref_reader:
            key = line.strip().split()[0]
            value = line.strip().split()[1:]
            if cn_postprocess:
                value = " ".join(value)
                value = value.replace(" ", "")
                value = [x for x in value]
            ref_dict[key] = value

    cer_detail_writer = open(cer_file, "w")
    for hyp_key in hyp_dict:
        if hyp_key in ref_dict:
            out_item = compute_wer_by_line(hyp_dict[hyp_key], ref_dict[hyp_key])
            rst["Wrd"] += out_item["nwords"]
            rst["Corr"] += out_item["cor"]
            rst["wrong_words"] += out_item["wrong"]
            rst["Ins"] += out_item["ins"]