2020/07/16

[yuseungwoo]

• wav 44100 -> wav 16000

  7018711@deploy-7018711-0646fb05ec584513b032a2811e8c22d8-c6957d69d-jtsft:/speech/SpeechCorpus/SpeechDB.44K$ sox music_search.m.0037.u0172.wav -r 16000 -c 1 -b 16 conv.wav

• 16kHz, spoken.query 다른 이유.. apple? 포맷 뭐지....

[yuseungwoo]

• soundIT DB 작업

cp -r /script/{scp,text}* list/

for i in ls *.scp;do iconv -c -f ASCII -t UTF-8 ${i} > utf8.${i};done for i in ls *.text;do iconv -c -f EUCKR -t UTF-8 ${i} > utf8.${i};done
for i in ls *.text;do iconv -c -f MSCP949 -t UTF-8 ${i} > utf8.${i};done
****mic.sent 등

Non ISO ASCII extended file => CP949, MSCP949, UHC, ... iconv 로 가능

dos2unix utf8.*

for i in ls ../;do cat utf8.${i}.text | awk -F"[0-9] " '{print $2}' | paste utf8.${i}.scp -| awk -F"[0-9] " '{print $2}';done for i in ls ../;do cat utf8.${i}.text | awk -F"[0-9] " '{print $2}' | paste utf8.${i}.scp -| awk -F"[0-9] " '{print $2}';done > list

내 수정본 기준 44.1kHz 31건 부족, 16kHz 8건 부족, 다른 이유.. apple? 포맷 뭐지....

[yuseungwoo]

차주 계획, 전사 enhance, raw 각각 500개 확보 목표