Question

我正在尝试使用SpeechRecognition转录大约3分钟的音频文件，但是，似乎无法转录超过20秒的任何内容。这是我正在使用的代码：

r = sr.Recognizer()

audio = FLAC(output_name +'.' + output_format)
audio_length = audio.info.length

file = sr.AudioFile(output_name +'.' + output_format)

with file as source:
    audio = r.record(source, duration = 20)

google = r.recognize_google(audio, language = 'ru-RU' )
print(google)

我该如何循环播放，使其转录0s-20s，然后转录20s-40s，依此类推，直到音频文件结束？

我想避免将文件尽可能地分成20s长度的单独文件。

Answer 1

所以我知道了。我的缺点是没有足够仔细地阅读SpeechRecognition模块的文档，但是它们有一个offset参数！

count = 0
for audio_path in audio_files:
     audio = FLAC(audio_list[count] + '.' + output_format) #specify audio file for length calculation
     audio_length = audio.info.length #get length of audio file

     #n.b. mutagen module used for calculating audio length

     number_of_iterations = int(audio_length/20)

    if number_of_iterations == 0:
        number_of_iterations = 1

     file = sr.AudioFile(audio_list[count] + '.' + output_format)


    for i in range(number_of_iterations):
        with file as source:
            audio = r.record(source, offset = i*20, duration = 20)

         google = r.recognize_google(audio, language = 'ru-RU' )
         count = count + 1
         print(google)

如何在20秒的间隔内遍历音频文件？

1 个答案: