Extending the E-Model Towards Super-Wideband and Fullband Speech Communication Scenarios
Extending the E-Model Towards Super-Wideband and Fullband Speech Communication Scenarios
复制标题
将E模型扩展到超宽带和全带语音通信场景
DOI：
10.21437/interspeech.2019-1340
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Hitoshi Aoki
中科院分区：
文献类型：
--
作者：
Sebastian Möller;Gabriel Mittag;Thilo Michael;Vincent Barriac;Hitoshi Aoki
为了根据用户体验的质量来规划语音通信服务，长期以来一直使用参数模型。这些模型基于描述传输信道和终端设备的元件的参数来预测通信伙伴所经历的总体质量。最常用的模型是ITU-T Rec中标准化的E-模型。G.107用于窄带和ITU-T记录。G.107.1用于宽带方案。然而，随着超宽带和全频带传输的到来，E-模型需要扩展。在这篇文章中，我们提出了一个扩展的E模型的第一个版本，它同时考虑了超宽带和全频带场景，并预测了语音编解码器、分组丢失和延迟的影响作为在这些场景中预期的最重要的降级。预测结果与纯听力测试和会话测试的结果以及基于信号的预测结果进行了比较，显示出合理的预测准确性。
In order to plan speech communication services regarding the quality experienced by their users, parametric models have been used since a long time. These models predict the overall quality experienced by a communication partner on the basis of parameters describing the elements of the transmission channel and the terminal equipment. The mostly used model is the E-model which is standardized in ITU-T Rec. G.107 for narrowband and in ITU-T Rec. G.107.1 for wideband scenarios. However, with the advent of super-wideband and fullband transmission, the E-model needs to be extended. In this paper, we propose a first version of an extended E-model which addresses both super-wideband and fullband scenarios, and which predicts the effects of speech codecs, packet loss, and delay as the most important degradations to be expected in such scenarios. Predictions are compared to the results of listening-only and conversational tests as well as to signal-based predictions, showing a reasonable prediction accuracy.