Fact · Dictation
8distances from the source
Prasad and Sim built a far-field speech corpus by re-recording Wall Street Journal audio with microphones at eight different distances from the source.
Checked against its source on October 16, 2026
In the source's own words
“A speech corpus was created by re-recording the Wall Street Journal (WSJ0) audio using far-field microphones with 8 different distances from the source.”
Microphone Distance Adaptation Using Cluster Adaptive Training for Robust Far Field Speech Recognition, Animesh Prasad and Khe Chai Sim, Interspeech 2016, International Speech Communication Association (ISCA), 2016, Abstract.
Open the original- What it measures
- The paper's corpus was built to test how a recognizer copes when the speaker's distance from one microphone changes. The distances are the study's design, not a measured WER curve.
- What it does not show
- It shows that distance was treated as the main variable. The paper uses far-field microphones in a room, not a headset, and it reports adaptation results, not a word-error-rate table by distance in the abstract.