教你3分鐘做出一個AI影片轉字幕程式！無需安裝軟硬體，google雲端輕鬆搞定

[@pitv1]

Colab 範例程式：
colab.research.google.com/drive/1daS9jIBivCuK7o_ikaau3CMQK2TwSiWv?usp=drive_link
前置作業：
!pip install faster_whisper nvidia-cublas-cu12
參考程式碼：
from faster_whisper import WhisperModel
from google.colab import drive
drive.mount('google硬碟')
trans = ""
model = WhisperModel("large-v3", device="cuda", compute_type="float16")
# 改成你的檔案名稱
filename = "google硬碟/MyDrive/test.mp3"
segs, info = model.transcribe(filename, beam_size=5, initial_prompt="臺灣的繁體中文")
for i, seg in enumerate(segs, 1):
startHH, startRR = divmod(seg.start, 3600)
startMM, startSR = divmod(startRR*1000, 60000)
startSI, startSF = divmod(startSR, 1000)
endHH, endRR = divmod(seg.end, 3600)
endMM, endSR = divmod(endRR*1000, 60000)
endSI, endSF = divmod(endSR, 1000)
trans += "%d
%02d:%02d:%02d,%03d --> %02d:%02d:%02d,%03d
%s

" % (
i,
startHH, startMM, startSI, startSF,
endHH, endMM, endSI, endSF,
seg.text
)
print (trans)
srtfile=open(filename+".txt", "w")
srtfile.write(trans)
srtfile.close()