Project description: Speech Corpus for Speech Recognition Assessment

---

◦ Recordings were made through high sensitive measurement microphones in 
a soundproof room to be used as a basic speech corpus for creating simulated 
speech data reflecting a variety of noise environments. The prompting items 
were composed of phonetically balanced words and sentences containing 5,000 
words of high frequency. One speaker pronounced 202 ~ 206 tokens
◦ A total of 300 speakers (Male: 150, Female: 150)
◦ Soundproof room
◦ Sampling and data format
 - 48,000 Hz, 16 bit Windows wave format
◦ Microphone
 - primary microphone: B&K 4189 (Prepolarized Free-Field 1/2")
 - secondary microphones:
◦ Prompts
 - 452 phonetically balanced words
 - 5 sentences containing all the Korean phonemes (for all the speakers)
 - 8,608 sentences containing 5,000 words of high frequency
◦ Per speaker
 - 5 sentences
 - 150 -152 phonetically balanced words 
 - 47-49 sentences of high frequency