
ããã«ã¡ã¯ãSHIRAN.CODERã§ãïŒ
ãã®ããŒãžã§ã¯ãColabã§äœæ¥ãé²ãããšãã«åå¿è ããã ãããïŒåããªãïŒïŒã ãšãªããã¡ãªãã€ã³ããšããµã¯ããšè§£æ±ºããæ¹æ³ããŸãšããŠãããŸãã
ç®æ¬¡[衚瀺]
1. ãã¡ã€ã«ã®ã¢ããããŒãåããºã¬ãåé¡
ããããçç¶
- åããã¡ã€ã«ã2åã¢ãããããš
session_transcript.txt
âsession_transcript (2).txt
ã«ãªã - ã³ãŒãã§ã¯
session_transcript.txt
ãèªãããã«ããŠããã®ã§ãšã©ãŒã«ãªã
ããã§è§£æ±ºïŒ
- å·ŠåŽã®ãã©ã«ãã¢ã€ã³ã³ïŒ
ïŒãéã
- ãããªããã¡ã€ã«ãå³ã®ãïžãããåé€ãã
- å¿ èŠãªãã¢ããããŒããçŽã
ãã€ã³ãïŒãã¡ã€ã«åãšã³ãŒãã®æå®ãäžèŽããŠããããå¿
ã確èªãããïŒ
2. ãclient is not definedããšã©ãŒ
ããããçç¶
client.chat.completions.create()
ãåŒãã ãNameError: name 'client' is not defined
ãåºã
ããã§è§£æ±ºïŒ
from openai import OpenAI
#
å¿
ãããã§ client ãäœã£ãŠããïŒ
client = OpenAI()
ãã€ã³ãïŒ
client
ã¯ãOpenAIãšäŒè©±ããããã®åä»ä¿ãã¿ãããªãã®ã
ãããäœããªããšãGPTã«äœãäŒãããŸããã
3. ããã¹ããå
šéš1è¡ã«ãªã£ãŠãã
ããããçç¶
- Whisperã§æåèµ·ããããããã¹ããæ¹è¡ãªãã§ããŒã³ãš1è¡
readlines()
ã§ã¯çºè©±ãåå²ã§ããªã
ããã§è§£æ±ºïŒ
import re
# ãã¡ã€ã«ãäžžããšèªã¿èŸŒã
raw_text = open("session_transcript.txt", "r").read()
# ããããïŒããªã©ã§åå²
segments = re.split(r"(ã|ïŒ|\?)", raw_text)
lines = [a + b for a, b in zip(segments[::2], segments[1::2])]
segments = [line.strip() for line in lines if line.strip()]
print(segments[:5]) # ã詊ã衚瀺
4. ãŸãã¯å°ãã ããã¹ãããæ¹æ³
ããããçå
- ãããªãå šéšã®çºè©±ãåŠçãããšæéãããã
- éäžã§ãšã©ãŒãèµ·ãããé¢åâŠ
ããã§è§£æ±ºïŒ
# æåã®10ä»¶ã ã詊ã
for i, seg in enumerate(segments[:10]):
...
ãã€ã³ãïŒ
segments[:10]
ã¯ãæåã®10ä»¶ã ã䜿ãããšããæå³ã
ãã¹ããçµãã£ãããsegments
ã«æ»ããŠå
šéšå®è¡ãããïŒ
5. Colabã®åèµ·å
ããããçç¶
- å€ãªãšã©ãŒãçŽããªã
- ãªããåãããããã
ããã§è§£æ±ºïŒ
- äžã¡ãã¥ãŒã®
[ã©ã³ã¿ã€ã ] â [ã©ã³ã¿ã€ã ãåèµ·å]
ãã¯ãªã㯠- ããäžåºŠãå¿ èŠãªã»ã«ãäžããé ã«å®è¡ãçŽã
ãŸãšã
ããã ãèŠããŠããã°ãColabã§ãåããªãïŒïŒããšãªã£ãŠãå®å¿ã§ãã
å¿
èŠãªãšãã¯ããã®ããŒãžãããã¯ããŒã¯ããŠãããŠãã ãããâ
é¢é£ãªã³ã¯
Colabã§æ¥æ¬èªã°ã©ããè±è
or ⊠ã«ãªãåå ãšå¯Ÿç
ãåå ãŸãšãã
Colabç°å¢ã«ã¯æ¥æ¬èªãã©ã³ããå
¥ã£ãŠããªãããšãå€ã
ã å¿
èŠãªãã©ã³ããæåã§
apt-get
ã§å
¥ããå¿
èŠããã
Matplotlib ã¯ãã©ã³ãåã ãã ãšããŸãèªèã§ããªãããšããã
ã ãã¡ã€ã«ãã¹ãçŽæ¥æå®ããŠ
FontProperties
ã䜿ããªããšå€±æããããã
Matplotlib ã¯è±èªäžå¿èšèšãªã®ã§æ¥æ¬èªçµçãèŠæ
ã ã©ãã«ãã¿ã€ãã«ãé·ããšåæã«ã...ãã§çç¥è¡šç€ºãããã
ãã·ã³ãã«ãªè§£æ±ºçã
ããã ãã§å
šéšè§£æ±ºïŒ
!apt-get -y install fonts-noto-cjk
!pip install japanize-matplotlib
import japanize_matplotlib # â ããããã€ã³ãïŒ
ãããå ¥ããã°ããã©ã³ãã®æåèšå®ã»ãã¹æå®ã»ãã£ãã·ã¥åæ§ç¯âŠå šéšãããªããŠOKã
title
, xlabel
, ylabel
ãæ®éã«æ¥æ¬èªã§æžãã ãã§è±è
ç¥ããïŒ
ãã³ããŸãšãã
Colabã§æ¥æ¬èªã°ã©ãã«ããæçã«ãŒã
â japanize-matplotlib
ãã€ã³ã¹ããŒã«ã㊠import
ããã ã
ã©ãããŠãæåã§ãããªã
â FontProperties
ã« .ttc
ãã¡ã€ã«ãã¹ãæž¡ã
æ¥æ¬èªããâŠãã«çç¥ãããã®ãå«ãªã
â japanize-matplotlib
ãäžçªæè»œã§ç¢ºå®