SpokesBiz
SpokesBiz is a freely available corpus of spontaneous conversational Polish, comprising over 600 hours of recordings and their transcriptions.
TryHow to use this tool?
1
A diverse corpus of authentic conversational Polish, which consists of 8 distinct sub-corpora varying in genre, size, and level of formality.
2
Each recording in the corpus was automatically transcribed with the use of a combination of ASR systems.The automatic transcripts were manually corrected and punctuated in accordance with the guidelines developed for the DiaBiz corpus.
3
SpokesBiz is currently available free of charge for non-commercial purposes. The instructions for accessing SpokesBiz are available here.
Features of the tool
Product feature
SpokesBiz comprises 8 distinct subcorpora, including a collection of Internet podcasts, biographical interviews, student presentations on academic and popular topics, thematic discussions, spontaneous conversations and more.
Product feature
A total of 504 speakers from different regions of Poland, with different educational backgrounds (from primary education to doctoral degrees) and of different ages participated in the formation of the corpus. All these factors contribute to the high diversity and representativeness of the corpus.
Product feature
The corpus data were automatically transcribed and time-aligned and subsequently manually corrected. The transcripts are also punctuated using guidelines developed for the DiaBiz corpus.
Product feature
The diversity of data in the corpus allows for conducting a variety of sophisticated linguistic analyses. SpokesBiz can be also successfully used in fields such as language teaching, translation studies, lexicography or language processing.
Product feature
The corpus contains a total of more than 650 hours of recordings, of which 600 hours have been transcribed. This in turn yields 5,911,420 words and 347,279 sentences.
Lorem ipsum dolor sit amet consectetur.
Lorem ipsum dolor sit amet consectetur. Lorem ipsum dolor sit amet consectetur. Lorem ipsum dolor sit amet consectetur. Lorem ipsum dolor sit amet consectetur.
Try