| 12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152 |
- from difflib import SequenceMatcher
- '''
- 1. two array, 1.Ground truth 2.from pytrancriber
- 2. get 1st index from both, if Ground Truth one is too short, get 2 or 3 even more
- 3.
- compare by mp3 len????
- '''
- def similar(a, b):
- return SequenceMatcher(None, a, b).ratio()
- def gen_mp3_from_text(gt,gen):
- tts=gTTS(text=gen, lang='zh')
- tts.save("gen_tmp.mp3")
- tts=gTTS(text=gt, lang='zh')
- tts.save("gt_tmp.mp3")
- audio = MP3("gen_tmp.mp3")
- gen_len = audio.info.length
- audio = MP3("gt_tmp.mp3")
- gt_len = audio.info.length
- return gt_len, gen_len
- def adjustSub(gt_array, generated_array):
- for gen_idx in
-
- return adjusted_array
- from gtts import gTTS
- tts=gTTS(text='玉米燕麥豆是一位居住在英國倫敦的日益女性', lang='zh')
- tts.save("1.mp3")
- tts=gTTS(text='yumi yamato是一位居住在英國倫敦的日益女性', lang='zh')
- tts.save("2.mp3")
- from mutagen.mp3 import MP3
- audio = MP3("1.mp3")
- print(audio.info.length)
- audio = MP3("2.mp3")
- print(audio.info.length)
|
