DESPERATELY SEEKING EMOTIONS OR: ACTORS, WIZARDS, AND HUMAN BEINGS.

DESPERATELY SEEKING EMOTIONS OR: ACTORS, WIZARDS, AND HUMAN BEINGS.
复制标题

拼命寻求情感或：演员、巫师和人类。

DOI：
--
复制
发表时间：
2000
期刊：
影响因子：
--
通讯作者：
J. Spilker
J. Spilker
中科院分区：
--
文献类型：
--
作者：
A. Batliner;K. Fischer;R. Huber;J. Spilker

文献摘要

被引文献

相似文献

例如，呼叫中心使用的自动对话系统应该能够在对话的关键阶段（由客户愤怒/恼怒的声音表达指示）确定何时最好将其传递给人工操作员。乍一看，这似乎不是一个复杂的任务：据文献报道，情感可以根据韵律特征相当可靠地区分开来。然而，这些结果大多数时候是在实验室环境中实现的，有经验的发言者（演员），并与引发，控制语音。我们报告了在不同的实验环境中使用韵律特征向量对两类问题“中性与愤怒”进行分类的结果，并讨论了单一特征对分类率的影响。这些设置的识别率对于特定于说话者的classi（cid：12）er（一个有经验的说话者，表演）是最好的，对于说话者独立的classi（cid：12）er（几个经验较少的说话者，阅读）是更差的，并且对于说话者独立的classi（cid：12）er（其中天真的受试者在绿野仙踪场景中执行约会调度的任务，其中模拟故障系统以唤起愤怒）甚至更差。（cid：12）第一种情况反映了文献中报道的大多数设置，第三种情况最接近“现实生活”任务
Automatic dialogue systems used in call-centers, for instance, should be able to determine in a critical phase of the dialogue - indicated by the costumers vocal expression of anger/irritation - when it is better to pass over to a human operator. Ata (cid:12)rst glance, thisseems not to be a complicated task: It is reported in the literature that emotions can be told apart quite reliablyon the basis of prosodic features. However, these results are most of the time achieved in a laboratory setting, with experienced speakers (actors), and with elicited, controlled speech. We report classi(cid:12)cation results obtained within di(cid:11)erent experimental settings for the two-class-problem ‘neutral vs. anger’ using a vector of prosodic features and discuss the impact of single features on the classi(cid:12)cation rate. Recognition rates for these settings are best for a speaker-speci(cid:12)c classi(cid:12)er (one experienced speaker, acting), worse for a speaker-independent classi(cid:12)er (several less experienced speakers, reading), and even worse for a speaker-independent classi(cid:12)er with naive subjects performing the task of appointment scheduling in a Wizard-of-Oz-scenario where a malfunctioning system is simulated in order to evoke anger. The (cid:12)rst situation mirrors most of the settings reported in the literature, the third is closest to the ‘real-life’-task