In multilingual speech recognition systems, a situation can often arise when the language is not known in advance, but the signal has already been received and is being processed. For such cases, some generalized model is needed that will be able to respond to phonetic differences and, depending on them, correctly recog-nize speech in the desired language. To build such a model, it is necessary to set the values of phonetic parameters, and then compare similar sounds, establishing significant differences.

翻译：在多语种语音识别系统中，常会出现语言未知但信号已接收并正在处理的情况。针对此类场景，需要构建某种通用模型，该模型能够感知语音差异，并据此正确识别目标语言的语音。要构建此类模型，需设定语音参数值，进而比较相似音素，确立显著差异。