Acoustic database of questions

As a result of participating in the Cross-Language Evaluation Forum (CLEF), it was created an acoustic database of questions in different languages, as Spanish, Arabic or Thai. The main goal is the use of these questions to train some speech recognition systems.

Methodology

• Recording of questions



• Recording file format: simple rate, 16.000Hz. Mono.
• Number of speakers: 10 (5 male y 5 female). Total number of questions: 500 (each speaker read 200 questions). Total number of sound files: 2000.
• Microphones: headphone with microphone Plantronics Audio 300 and microphone Plantronics Audio 330.

Consult online