bugfix v1.1.11

游雁

2024-10-10 2330e58f5ffa6d78cbdc099ac1a7e0aa1bd6a09f

bugfix v1.1.11

 funasr/utils/postprocess_utils.py

@@ -126,10 +126,10 @@
                end = time_stamp[ts_nums[num]][1]
                ts_lists.append([begin, end])
        else:            
            word_lists.append(words[num])
            # length of time_stamp may not equal to length of words because of the (somehow improper) threshold set in timestamp_tools.py line 46, e.g., length of time_stamp can be zero but length of words is not.
            # Moreover, move "word_lists.append(words[num])" into if clause, to keep length of word_lists and length of ts_lists equal.
            if time_stamp is not None and ts_nums[num]<len(time_stamp) and words[num] != " ":
                word_lists.append(words[num])
                begin = time_stamp[ts_nums[num]][0]
                end = time_stamp[ts_nums[num]][1]
                ts_lists.append([begin, end])
@@ -302,6 +302,7 @@
    sentence = "".join(word_lists)
    return sentence, real_word_lists


emo_dict = {
        "<|HAPPY|>": "😊",
        "<|SAD|>": "😔",
@@ -365,7 +366,15 @@
}

emo_set = {"😊", "😔", "😡", "😰", "🤢", "😮"}
event_set = {"🎼", "👏", "😀", "😭", "🤧", "😷",}
event_set = {
    "🎼",
    "👏",
    "😀",
    "😭",
    "🤧",
    "😷",
}


def format_str_v2(s):
        sptk_dict = {}
@@ -386,9 +395,11 @@
                s = s.replace(emoji + " ", emoji)
        return s.strip()


def rich_transcription_postprocess(s):
        def get_emo(s):
                return s[-1] if s[-1] in emo_set else None

        def get_event(s):
                return s[0] if s[0] in event_set else None

@@ -410,4 +421,3 @@
                new_s += s_list[i].strip().lstrip()
        new_s = new_s.replace("The.", " ")
        return new_s.strip()
    

 funasr/version.txt

@@ -1 +1 @@
1.1.10
1.1.11