Question

我试图遍历文件中的句子，选择“最佳”句子（这是稀有双音素（声音）数量最多的句子），并在选择了句子之后，更改的字典值句子中的每个diphone都设为0，这样就不会再次选择diphone（因为我要确保已选择所有可能的diphone）。

我已经为此编写了代码，但是看不到为什么它不影响输出，因为当我检查在for循环开始时选择的其中一个字典键的值时，已将其设置为0。我的代码是：

diphone_frequencies = {...}
diphone_frequencies_original = copy.deepcopy(diphone_frequencies)

line_score = {}
best_utts = []

for i in range(650):
    # Open the file and put all its lines in a list. Once.
    with open('recipe_diphone_utts.txt') as file:

        # Read the file lines one by one.
        for line in file:
            line = line.rstrip('\r\n')
            if line in best_utts:
                continue # Skip previously picked sentences.
            score = 0.0
            # Add a score to the line depending on its content.
            for word in line.split():
                score += float(diphone_frequencies[word])
            line_score[line] = score/len(line.split())

        # Sort each lines based on their score and get the best.
        best_sentence = max(line_score.keys(), key=(lambda k: line_score[k]))
        best_utts.append(best_sentence)
        print(best_sentence)

        # Each unique word of this iteration's best sentence has its score set to 0.
        for item in set(best_sentence.split()):
            diphone_frequencies[item] = 0

        if all(value == 0 for value in diphone_frequencies.values()):
            diphone_frequencies = diphone_frequencies_original

编辑：这已解决，但是我现在无法接受自己的回答；问题是在打开文档后出现了for循环；当我放

时，代码正常工作

for i in range(600):

之前

with open('recipe_diphone_utts.txt') as file:

编辑2：

面临的主要问题已解决，我已经更改了代码，但是行：

if line in best_utts:
    continue

应该确保一旦重置字典值就不会再次选择同一行的多个实例，但是这会导致一遍又一遍地将同一句子选为最佳句子，因此我需要采取其他方法来防止同一句话被多次选择。

Answer 1

当前best_utts == [best_sentence] * 600由于存在外部循环，与文件中所有其他句子（行）相比，best_sentence是得分最高的句子。

要获得600句最好的句子，我会这样：

diphone_frequencies = {...}
diphone_frequencies_original = copy.deepcopy(diphone_frequencies)

line_score = {}
best_utts = []

# Open the file and put all its lines in a list. Once. 
with open('recipe_diphone_utts.txt') as file:
    all_lines = file.readlines()

for i in range(600):
    print(diphone_frequencies['f_@@r'])

    # Read the file lines one by one.
    for line in all_lines:
        line = line.rstrip()
        if line in best_utts:
            line_score[line] = 0
            continue # Skip previously picked sentences.
        score = 0.0
        # Add a score to the line depending on its content.
        for word in line.split():
            score += float(diphone_frequencies[word])
        line_score[line] = score/len(line.split())

    # Sort each lines based on their score and get the best.
    best_sentence = max(line_score.keys(), key=(lambda k: line_score[k]))
    best_utts.append(best_sentence)

    # Each unique word of this iteration's best sentence has its score set to 0.
    for item in set(best_sentence.split()):
        diphone_frequencies[item] = 0

    if all(value == 0 for value in diphone_frequencies.values()):
        diphone_frequencies = diphone_frequencies_original

print(best_utts)

由于您使用的是file.close()而不是with open ... as file，所以最后也不需要file = open(...)。

Answer 2

我发现我犯下的主要错误是

for i in range(600):

之后

with open('recipe_diphone_utts.txt') as file

当我将其更改为with with ...在for循环内时，它起作用了。

用于循环更新字典值但不影响输出

2 个答案: