search

kaldi-asr / kaldi

kaldi-asr/kaldi is the official location of the Kaldi project.
http://kaldi-asr.org
Other
14.03k stars 5.3k forks source link

Does Kaldi fail to extract features from the non-standard pronunciation of ethnic minorities? #4871

Open LijingDK opened 11 months ago

LijingDK commented 11 months ago

steps/make_mfcc_pitch.sh --nj 64 --mfcc-config conf/mfcc_hires.conf --cmd run.pl data/test utils/validate_data_dir.sh: Successfully validated data-directory data/test steps/make_mfcc_pitch.sh: [info]: no segments file exists: assuming wav.scp indexed by utterance. steps/make_mfcc_pitch.sh: It seems not all of the feature files were successfully procesed (4379 != 4596); consider using utils/fix_data_dir.sh data/test steps/make_mfcc_pitch.sh: Succeeded creating MFCC and pitch features for test steps/compute_cmvn_stats.sh data/test Succeeded creating CMVN stats for test fix_data_dir.sh: kept 4379 utterances out of 4596 fix_data_dir.sh: old files are kept in data/test/.backup

LijingDK commented 11 months ago

Every time the number of nj is modified, the number of features that can be generated is also constantly changing. Different numbers of nj have different number of features. Why is this? The audio format is not corrupted

danpovey commented 9 months ago

That command would have created log files that would have warnings about any problems, use 

 find . -name '*.log' -mtime -2 --print

as an example command to find such files. I don't know how you think it's possible that Kaldi would treat recordings of ethnic minorities differently than the ethnic majority-- how would it know? These features just relate to the frequency spectrum.

LijingDK commented 9 months ago

Hello, I'm glad you can answer my question, my problem has now been solved, not the problem of kaldi, nor the problem of the pronunciation quality of the ethnic minorities, the main thing is that my two audios (two words) are problematic, which led to this problem, after solving this problem, all the characteristics are solved.

------------------ 原始邮件 ------------------ 发件人: "Daniel @.>; 发送时间: 2023年11月10日(星期五) 下午4:29 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [kaldi-asr/kaldi] Does Kaldi fail to extract features from the non-standard pronunciation of ethnic minorities? (Issue #4871)

That command would have created log files that would have warnings about any problems, use find . -name '*.log' -mtime -2 --print
as an example command to find such files. I don't know how you think it's possible that Kaldi would treat recordings of ethnic minorities differently than the ethnic majority-- how would it know? These features just relate to the frequency spectrum.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>