Python中Google语音识别失败回调chunks_process后for循环重复执行问题求解

问题根因 #

你的代码存在两个核心问题导致重复执行：

• 递归调用chunks_process处理完从i到末尾的所有分片后，当前层的for循环没有终止，会继续从i+1遍历后续已经被递归处理过的分片，产生重复执行
• Python中字符串是不可变类型，递归内部对whole_text的修改不会同步到外层函数，最终拼接结果会丢失递归处理的内容
• 额外风险：裸except会捕获所有异常包括键盘中断等非接口错误，且没有重试次数限制，极端情况会触发栈溢出

最优修复方案（非递归重试） #

直接在当前循环内对识别失败的分片做有限次重试，逻辑更清晰无重复执行风险：

import os
import speech_recognition as sr

def chunks_process(chunks, folder_name, max_retry=3):
    r = sr.Recognizer()
    whole_text = ""
    for i in range(len(chunks)):
        chunk_filename = os.path.join(folder_name, f"chunk{i}.wav")
        chunks[i].export(chunk_filename, format="wav")
        with sr.AudioFile(chunk_filename) as source:
            audio_listened = r.record(source)
            # 单分片内重试最多max_retry次
            text = None
            for retry_cnt in range(max_retry):
                try:
                    text = r.recognize_google(audio_listened, language='vi-VN')
                    break
                except sr.UnknownValueError:
                    print(f"分片{i}识别失败，重试第{retry_cnt+1}次")
                except sr.RequestError as e:
                    print(f"接口请求失败：{e}，重试第{retry_cnt+1}次")
            if text:
                formatted_text = f"{text.capitalize()}. "
                print(chunk_filename, ":", formatted_text)
                whole_text += formatted_text
            else:
                print(f"分片{i}重试{max_retry}次仍失败，跳过")
    return whole_text

递归版本修复（如果必须保留递归实现） #

去掉原函数内的for循环，完全通过pos参数控制处理进度，避免重复执行：

def chunks_process(chunks, pos, folder_name, whole_text, max_retry=3, current_retry=0):
    # 所有分片处理完成返回结果
    if pos >= len(chunks):
        return whole_text
    # 当前分片重试次数耗尽，跳过处理下一个
    if current_retry >= max_retry:
        return chunks_process(chunks, pos+1, folder_name, whole_text, max_retry, 0)
    r = sr.Recognizer()
    chunk_filename = os.path.join(folder_name, f"chunk{pos}.wav")
    chunks[pos].export(chunk_filename, format="wav")
    with sr.AudioFile(chunk_filename) as source:
        audio_listened = r.record(source)
        try:
            text = r.recognize_google(audio_listened, language='vi-VN')
        except (sr.UnknownValueError, sr.RequestError):
            # 识别失败重试当前分片
            return chunks_process(chunks, pos, folder_name, whole_text, max_retry, current_retry+1)
        else:
            text = f"{text.capitalize()}. "
            print(chunk_filename, ":", text)
            whole_text += text
            # 处理下一个分片
            return chunks_process(chunks, pos+1, folder_name, whole_text, max_retry, 0)

调用时初始传入pos=0、whole_text=""即可。

内容的提问来源于stack exchange，提问作者Ochimot