Speech enhancement is used to improve speech quality by using various algorithms. In Speech Enhancement, speech degraded by noise, or noise reduction, is the most important field and used for many applications such as mobile phones, VoIP, teleconferencing systems, speech recognition, and hearing aids.
Picus Tech portfolio of speech enhancement
Acoustic Echo Canceller (AEC)
View Details
Acoustic Echo Canceller (AEC)
Echo cancellation eliminates echo in a two-way voice transmission. Picus Tech implementation of AEC uses subband filter approach compensating echo in each sub band. The approach allows for superior cancellation. The echo cancellation as well as adaptation continues during double talk greatly enhancing the user experience.
Applications
- Video Conference
- Hands-free car phone systems
- Mobile devices
Supported Platforms
- Cortex 32/64 bit A Series - A8, A35, A53, A73, Neon
- Cortex M Series - M3, M4, M7, M33
Picus Tech Implementation
- Fast rate of adaptation
- Superior audio quality, preserves naturalness of speech after cancellation
- Robust double talk detection.
- High convergence
- Low delay
- Supports Variable Tail length
- Adapts to change in echo path
- AEC parameters can be configured
- Optimized for target platform
- Can be easily integrated on Android or Linux platforms.
- Superior echo cancellation without introducing audio artifacts or discontinuities
- Multi-channel support
- Supported sampling rates are 8khz, 11.025khz, 12khz, 16khz, 22.05khz, 24khz, 32khz, 44.1khz, 48khz
Noise Reduction
View Details
Noise Reduction
Noise reduction (Noise Suppression) module eliminates unwanted noise from the signal by determining actual noise level in the speech itself.
Applications
- Video Conferencing
- Hands-free car phone systems
- Mobile devices
Supported Platforms
- Cortex 32/64 bit A Series - A8, A35, A53, A73, Neon
- Cortex M Series - M3, M4, M7, M33
Picus Tech Implementation
- Superior Noise Reduction, preserves naturalness of speech
- Automatic threshold detection
- No fixed threshold required for VAD detection
- Low memory requirements
- Fully re-entrant implementation eases memory allocation
- Simple and easy to use API, which allows easy integration
- Highly optimized in assembly for best performance
- Easily extendible to other platforms
- Optimized for target platform
- Can be easily integrated on Android or Linux platforms.
Fast rate of adaptation