Audio Processing

voice INTER connect has extensive experience in developing relevant algorithms in the field of speech recognition and speech processing:

Acoustic echo cancellation (AEC)

• AEC according to ITU-T G.167
• Negligible rest echo, ERLE: approx. 40 dB
• Low memory requirements for data and programs, low system requirements
• Continuous adaptation and rapid convergence in real-time
• Reliable function even with small SNR
• Efficient integrated noise suppression for noisy environments
• Input of comfort noise

Noise reduction (NR)

• No distortion of voice characteristics
• 25 dB improvement of the SNR
• Continuous adaptation to background noise
• Low memory requirements and system requirements

Voice activity detection (VAD)

• Fast adaptation to changing environmental conditions
• Robust detection of low energetic speech shares, e.g. fricative
• Particularly appropriate for disturbed environments through combination with noise reduction
• Especially appropriate for implementation on a DSP
• Very good operation in noisy environments due to combination with algorithms of noise reduction