Voice Methods | Linguistic Databases

We've created these corpuses not for academic purposes, but for the development of business solutions. Any developer of speaker-independent Polish and/or Russian speech recognition systems may acquire them to help achieve their goals. The corpuses may be used to create automated call centers, voice-operated e-mail access systems, voice dialing systems and more.

Russian Acoustic-Phonetic Continuous Speech Corpus

Data type: speech
Number of speakers: 46
Number of spoken sentences: 150
Average number of phonemes in a sentence: 60
Total running time: 7.75 hours
Sample rate: 16000 Hz
Sampling format: 1 channel 16-bit linear
Data source(s): microphone
Recording conditions: office, quiet room
Application(s): speech recognition
Delivery format: CD's containing pcm-files
Total file size: 900 Mb

Polish Acoustic-Phonetic Continuous Speech Corpus

Data type: speech
Number of speakers: 100
Number of spoken sentences: 105
Average number of phonemes in a sentence: 35
Total running time: 9.8 hours
Sample rate: 16000 Hz
Sampling format: 1 channel 16-bit linear
Data source(s): microphone
Recording conditions: office, quiet room
Application(s): speech recognition
Delivery format: CD's containing pcm-files
Total file size: 1,13 Gb

Polish Acoustic-Phonetic Speech Corpus (Isolated Letters)

Data type: speech
Number of speakers: 100
Method of speaking: Each letter of the alphabet is pronounced twice separately and once within a reading of the entire alphabet
Total running time: 2.8 hours
Sample rate: 16000 Hz
Sampling format: 1 channel 16-bit linear
Data source(s): microphone
Recording conditions: office, quiet room
Application(s): speech recognition
Delivery format: CD's containing pcm-files
Delivery format: 320 Mb

To obtain more information about licensing and acquisition of our speech corpuses, contact us now.