KALDI之aishell之V1模型续4（最终的结果）

秒速五厘米 2022-04-17 02:08 268阅读 0赞

成功结束了aishell之V1模型：

eer=0.049%

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70][]

sid/extract_ivectors.sh --cmd run.pl --mem 4G --nj 10 exp/extractor_male data/test/enroll exp/ivector_enroll_1024
    sid/extract_ivectors.sh: extracting iVectors
    sid/extract_ivectors.sh: combining iVectors across jobs
    sid/extract_ivectors.sh: computing mean of iVectors for each speaker and length-normalizing
    sid/extract_ivectors.sh --cmd run.pl --mem 4G --nj 10 exp/extractor_male data/test/eval exp/ivector_eval_1024
    sid/extract_ivectors.sh: extracting iVectors
    sid/extract_ivectors.sh: combining iVectors across jobs
    sid/extract_ivectors.sh: computing mean of iVectors for each speaker and length-normalizing
    compute-eer -
    LOG (compute-eer[5.5.39~1-88f23]:main():compute-eer.cc:136) Equal error rate is 0.0487076%, at threshold -12.62450.04871

错误率太低，怀疑自己拿了训练数据当测试数据 ，然后尴尬成训练自己又识别自己吗

data下的feats.scp

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 1][]

test下的wav.scp

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 2][]

继续查看打分部分总脚本：

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 3][]

找run.sh中trials这个第一次出现

![20181111212252971.png][]

根据trails路径打开文件

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 4][]

把data/test的utt2spk打开看看

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 5][]

github源码中的错误率Equal error rate is 0.140528%, at threshold -12.018；

我的错误率 Equal error rate is 0.0487076%, at threshold -12.6245

同样的数据集：[http://www.openslr.org/resources/33/][http_www.openslr.org_resources_33]

![watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 6][]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70]: /images/20220417/60e056a1a0bf44dc892f1bfb53fb4d9c.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 1]: /images/20220417/a185968609f842d58d03604303a554f9.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 2]: /images/20220417/4da3773aad82489684fa35e42bafe918.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 3]: /images/20220417/4d5ff8a08895482f96823e0287cdbb2c.png
[20181111212252971.png]: /images/20220417/8b961a5c4b4748349a02d0ea2607e401.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 4]: /images/20220417/22748f2c22d7486f8c2cbf22f95027c1.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 5]: /images/20220417/0f267caaffb74c37ad37a9c1f032430e.png
[http_www.openslr.org_resources_33]: http://www.openslr.org/resources/33/
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dlaXhpbl8zODg1ODg2MA_size_16_color_FFFFFF_t_70 6]: /images/20220417/5f19578290094988afe5a01999b98967.png