The customer is the world leader in mobile communications from Finland, one of the largest telecommunication companies. Lanit-Tercom introduced very sophisticated algorithms for echo compensation, which isolates the echo from the proper signal and reduces it. Since these algorithms were absolutely new, several modeling levels were developed because it was not possible to predict their effectiveness.
Challenge: Musical and speech sounds reproduced in halls generate the echo signal, which can both attribute an elegant sound color or add a noisy disturbance depending on the musical genre and its purpose. The echo signal influence the sound reinforcement through resonance when a microphone is in the field of the loudspeaker and the acoustic feedback occurs.
There were the following problems to be solved:
- Compensation of a resonance and increase of the sound reinforcement gain
- On-line extraction and attenuation of the echo signal
- Restoration of the hall reverberation
- Correction of the reverberation level for the old records
- Investigation of the hall reverberation structure
The signal is divided into segments of the length 1-3 s, and the Fourier analysis is applied to this set.
Solution: The problems of the echo compensation and reverberation analysis were solved by the signal processing based on the Fourier analysis. The proprietary math algorithms were developed and tested. The approach was based on the precise analysis of the sound spectrum and estimation of the transfer function from the source sound to echo.