VOICE Awards Corpus of Human-Machine Communication
The VOICE Awards were an annual competition for German state-of-the-art commercially deployed spoken dialog systems (SDSs) that took place from 2005 to 2009. Participating SDSs came from virtually all available domains, such as games, banking, transportation, and telecommunications. Expert and novice testers performed typical tasks with each of the submitted SDSs and rated the systems’ quality according to a number of criteria.
The corpus consists of the recordings of all interactions between systems and testers, which were manually transcribed and highly detailed annotated by the SpeechEval research group. The annotations include task success, dialog acts, and communication problems. During transcription a rough segmentation into turns and dialog acts was performed. Overall, 1970 dialogs are recorded from 120 distinct systems, 17 of them entered in the VOICE Award contest in several consecutive years.
Available Data
| Data/Annotations |
|---|
| Audio |
| Transcription |
| Dialog acts |
| Error/Miscommunication |
| Repetitions |
| Dialog success |
| User ratings (per user/system) |
| Domain (per system) |
For more details visit the documentation of the VOICE Awards Corpus.