cer

游雁

2024-02-29 b5ea9c7a6a7cf0816fd59d7b3377752390d3a775

cer

 funasr/metrics/wer.py

File was renamed from funasr/metrics/compute_wer.py
@@ -1,10 +1,13 @@
import os
import numpy as np
import sys
import hydra

def compute_wer(ref_file,
                hyp_file,
                cer_detail_file):
                cer_file,
                cn_postprocess=False,
                ):
    rst = {
        'Wrd': 0,
        'Corr': 0,
@@ -24,14 +27,22 @@
        for line in hyp_reader:
            key = line.strip().split()[0]
            value = line.strip().split()[1:]
            if cn_postprocess:
                value = value.replace(" ", "")
                value = [x for x in value]
                value = " ".join(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 = value.replace(" ", "")
                value = [x for x in value]
                value = " ".join(value)
            ref_dict[key] = value

    cer_detail_writer = open(cer_detail_file, 'w')
    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])
@@ -47,6 +58,7 @@
           cer_detail_writer.write(hyp_key + print_cer_detail(out_item) + '\n')
           cer_detail_writer.write("ref:" + '\t' + " ".join(list(map(lambda x: x.lower(), ref_dict[hyp_key]))) + '\n')
           cer_detail_writer.write("hyp:" + '\t' + " ".join(list(map(lambda x: x.lower(), hyp_dict[hyp_key]))) + '\n')
           cer_detail_writer.flush()

    if rst['Wrd'] > 0:
        rst['Err'] = round(rst['wrong_words'] * 100 / rst['Wrd'], 2)
@@ -59,6 +71,7 @@
    cer_detail_writer.write("%SER " + str(rst['S.Err']) + " [ " + str(rst['wrong_sentences']) + " / " + str(rst['Snt']) + " ]" + '\n')
    cer_detail_writer.write("Scored " + str(len(hyp_dict)) + " sentences, " + str(len(hyp_dict) - rst['Snt']) + " not present in hyp." + '\n')

    cer_detail_writer.close()
     
def compute_wer_by_line(hyp,
                        ref):
@@ -146,12 +159,21 @@
            + str(rst['sub']) + ") corr:" + '{:.2%}'.format(rst['cor']/rst['nwords'])
            + ",cer:" + '{:.2%}'.format(rst['wrong']/rst['nwords']))

if __name__ == '__main__':
    if len(sys.argv) != 4:
        print("usage : python compute-wer.py test.ref test.hyp test.wer")
        sys.exit(0)

    ref_file = sys.argv[1]
    hyp_file = sys.argv[2]
    cer_detail_file = sys.argv[3]
    compute_wer(ref_file, hyp_file, cer_detail_file)
@hydra.main(config_name=None, version_base=None)
def main_hydra(cfg: DictConfig):
    ref_file = cfg.get("ref_file", None)
    hyp_file = cfg.get("hyp_file", None)
    cer_file = cfg.get("cer_file", None)
    cn_postprocess = cfg.get("cn_postprocess", False)
    if ref_file is None or hyp_file is None or cer_file is None:
        print("usage : python -m  funasr.metrics.wer ++ref_file=test.ref ++hyp_file=test.hyp ++cer_file=test.wer ++cn_postprocess=false")
        sys.exit(0)
        
    compute_wer(ref_file, hyp_file, cer_file, cn_postprocess)

if __name__ == '__main__':
    main_hydra()




 funasr/models/llm_asr_nar/model.py

@@ -315,7 +315,8 @@
        model_outputs = self.llm(inputs_embeds=inputs_embeds, attention_mask=attention_mask, labels=None)
        preds = torch.argmax(model_outputs.logits, -1)
        text = tokenizer.batch_decode(preds, add_special_tokens=False, skip_special_tokens=True)
        text = text[0].split(': \n')[-1]
        text = text[0].split(': ')[-1]
        text = text.strip()
        # preds = torch.argmax(model_outputs.logits, -1)
        
        ibest_writer = None