Python语言检测API是否支持多语言？如何实现语音转文本自动语种检测

核心思路 #

谷歌语音识别接口本身自带多语种检测能力，不需要额外引入第三方检测工具，仅需要调整recognize_google的调用参数，提取返回结果中自带的语种检测字段即可实现需求。

具体修改步骤 #

• 移除recognize_google方法中固定的language='ko'参数，新增show_all=True参数，获取接口返回的完整识别数据，数据中会包含匹配度最高的语种信息
• 从返回的完整数据中提取检测到的语种代码和对应识别文本
• 保留原有翻译和异常处理逻辑即可

修改后完整代码 #

import speech_recognition as sr
from googletrans import Translator
from textblob import TextBlob

translator = Translator()
speech = sr.Recognizer()

with sr.Microphone() as source:
    print("say something!")
    audio = speech.listen(source)
    try:
        # 开启show_all获取完整返回结果，不固定识别语种
        full_result = speech.recognize_google(audio, show_all=True)
        if not full_result:
            raise sr.UnknownValueError
        
        # 提取匹配度最高的识别结果和对应检测到的语种
        best_match = full_result['alternative'][0]
        detected_lang = full_result['language']
        ToTranslate = best_match['transcript']

        print("The language used is: " + detected_lang)
        print("You said: " + ToTranslate)
        translation = translator.translate(ToTranslate).text
        print("In english: " + translation)
    except sr.UnknownValueError:
        print("Sorry I can't understand you.")
    except sr.RequestError:
        print("Error connecting to speech recognition service.")

补充说明 #

如果需要限制可识别的语种范围，可以给recognize_google传入language_hints参数，传入优先匹配的语种代码列表，例如language_hints=['ko','en','zh-CN']，可提升指定语种的识别准确率。

内容的提问来源于stack exchange，提问作者Daniel Bueno