Speech Enhancement Algorithms for Array of Microphones

SEA²M™ (Speech Enhancement Algorithms for Array of Microphones*) is patented technology for hands-free communication that uses beam-forming array technology to focus a microphone beam on the individual speaker, virtually eliminating all ambient room noise and side conversations.

Usage of the hands-free speech communication systems implies unspecified talker position in the acoustic environment, with variable distances from system’s microphones and loudspeakers. Hands-free speech communication in such unknown conditions is the reason for number of technical problems, which should be solved, in order to preserve good quality of speech communication.

Typical Application: Hands-free videophone based on TV set with 5 microphone array and SEA2M™.

All technical solutions depend on multi-microphone sensor systems, and the difference between them is a specific processing of microphone signals1. SEA²M™ system offers one effective solution for hands-free communication systems and their usage in closed acoustic spaces of small and medium dimensions.

SEA²M™ can be applied in real-time systems such as:

• Video-phone systems
• Teleconference systems
• Room and car hands-free systems
• Human-machine voice interfaces, etc.

Overview

SEA²M™ solves inherent problems of hands-free full-duplex voice communication, like existence of loud acoustic echo, high noise level, room reverberation. Due to its adaptive nature, successfully handles dynamically changing conditions. It provides high quality conversation, and offers natural sound with high perspicuity.

SEA²M™ offers the following algorithms:

• Acoustic echo cancellation
• Determination of the speaker direction
• Audio beam-forming
• Post-processing
• Noise suppression
• Automated gain control

SEA²M™ is available as an ANSI C software IP package. It is portable and fully configurable. Due to its modular architecture, it can be easily integrated into numerous systems.

The parts of SEA²M™ are microphone array with 5 microphones and digital camera placed on the top of TV screen.

Thanks to SEA²M™ Speech Enhancement Algorithms for Array of Microphones system, the user can freely watch TV program and simultaneously communicate with remote party during his unobstructed moving in living room environment.

The demo offers comfort wireless communication through GSM or VoIP.

Features

SEA2M Acoustic Echo Canceller test

Global features of SEA2M solution:

• Signal processing is completely done in frequency domain
• Optimal system performances using common algorithmic resources
• Open for configuration changes with different numbers of microphones and loudspeakers
• Support specific application requirements with simple adaptation of pre-defined set of algorithmic parameters

Sound samples:

• Raw signal recorded by the microphone array
• Signal processed with SEA2M

Specifications

SEA2M Ambient Room

Acoustic Echo Canceling (AEC) technology:

• Echo canceling without post-processing**, up to 20dB

Acoustic beam-forming (BF) technology:

• Adaptive beamforming: 5 microphones
• Adaptive steering on speaker location
• Azimuth limits*: up to 20 dB

Noise Reduction (NR) technology

• Noise reduction capability: up to 20 dB

Automatic Gain Control (AGC) technology

Audio bandwidth

• 200Hz to 3.6kHz, sampling rate: 8kHz
• 200Hz to 5kHz, sampling rate: 11.025Hz

Echo Return-Loss Enhancement (ERLE)**

• Whole system: up to 28 dB

Sppech-to-Noise Ratio Enhancement (SNRE)**

• Whole system: up to 18 dB

Perceptual Evaluation of Speech Quality (PESQ)**

• Whole system: up to 2.9

DSP memory capacity: <120 kW

Computational Speed <200 MIPS

* Serbian patent applications: P-2006/0551, P-2006/0611, P-2006/0612, P-2006/0642
Serbian trademark registration: SEA²M Z-2281/06

**test room: Volume=105m3, Reverberation=330ms