Audio and Speech Solutions for Communications Equipment

The Noise Suppression (NS) Library is used to reduce unwanted noise mixed with speech. The NS Library is suitable for applications without an isolated noise reference source. The noise suppression algorithm is a frequency-domain algorithm based on spectral subtraction. A voice activity detector differentiates noise and speech. Noise reduction filters self adjust every 10 ms during periods of speech inactivity (using an FFT). The speech is continuously filtered, reducing its noise content

The Acoustic Echo Cancellation (AEC) Library is used to cancel unwanted echo caused by acoustic path from speaker to microphone. The AEC Library is ITU G.167 compliant and suitable for systems where  the speaker and microphone are close to each other. The library supports full-duplex operation and is based on Least Mean Square (LMS) adaptive filtering. A voice activity detector and double talk detector determine when the adaptive filter should be adjusted. An LMS filter iteratively models the echo propagation path during periods of far-end speech. The filtered output is subtracted from the microphone input every 10 ms, thus removing echo. The AEC Library supports cancellation of echo tail lengths that are 16, 32 or 64 ms long and can be used with audio signals with a bandwidth of 0–4 kHz at an 8 kHz sampling rate.

The Line Echo Cancellation (LEC) Library is used to cancel unwanted echo generated by various telephone circuits (e.g. hybrids) or digital network components. The LEC algorithm is a time-domain algorithm based on Least Mean Square (LMS) adaptive fltering. It is an ITU G.168-compliant solution and supports echo tail lengths that are 16, 32 or 64 ms long.

The dsPIC DSC Speech and Audio Fast Forward (SAFF) tool provides users with real-time control of Microchip’s Speech and Audio Processing Algorithms as they are running in an embedded system application. This PC GUI tool facilitates easy tuning of Microchip’s Noise Suppression, Acoustic Echo Cancellation, Line Echo Cancellation and Equalizer algorithms in speech and audio applications by simplifying parametric tuning of algorithms.

The dsPIC DSC Automatic Gain Control Library automatically adjusts the amplitude of a speech signal to match a set level. This is useful in speech applications where the distance between the speech source and the microphone is not fixed. The Automatic Gain Control Library can be used with our speech and audio solutions for speech signal pre-processing.