Acoustic Beamforming

ACOUSTIC BEAMFORMER SOFTWARE IMPROVES THE SIGNAL TO NOISE RATIO OF SPEECH SIGNALS

Microphone Array Beamforming

Multi-Mic

Applications include speech recognition, acoustic echo cancellation, and automotive “hands-free” audio input systems.

Features List (Preliminary)

  • Scalable number of microphones from 2 to 8
  • Supports fixed-angle or dynamic array steering
  • Programmable sampling rate
  • Programmable frame size
  • Programmable number of microphones
  • Programmable microphone spacing
  • Programmable microphone delay
  • Programmable microphone gain/loss
  • specify per microphone delay OR microphone to microphone spacing
  • Can be integrated with Adaptive Digital’s Acoustic Echo Canceller
  • Functions are C-callable
  • Multi-channel operation
  • C callable
  • Designed for half-wave microphone separation

Availability

Platforms
Texas Instruments – TI TMS320C6000 C67x
Texas Instruments – TI TMS320C5000 C55x , C54x

ADT Acoustic Beamformer is available on the above Platforms: Other configurations are available upon request.

Specifications

 

Acoustic Beamformer C67x

CPU UTILIZATION & MEMORY REQUIREMENTS
All Memory usage is given in units of byte.

Memory and MIPS data were recorded with an input frame size of 160 samples at an 8kHz-sampling rate.
VariantMIPS (Peak)Program MemoryData MemoryPer Channel Data MemoryScratch Memory
4 Microphones148473612431608192

Acoustic Beamformer C55x

CPU UTILIZATION & MEMORY REQUIREMENTS
All Memory usage is given in units of byte.

Memory and MIPS data were recorded with an input frame size of 160 samples at an 8kHz-sampling rate.
VariantMIPS (Peak)Program MemoryData MemoryPer Channel Data MemoryScratch Memory
2 Microphones2.4230033021800
4 Microphones4.8230033021800

Acoustic Beamformer C54x

CPU UTILIZATION & MEMORY REQUIREMENTS
All Memory usage is given in units of byte.

Memory and MIPS data were recorded with an input frame size of 160 samples at an 8kHz-sampling rate.
VariantMIPS (Peak)Program MemoryData MemorySNR Improvement
2 Microphones19.8194840613.1 dB
4 Microphones39.7194861876.2 dB

Description

The Adaptive Digital Acoustic Beamformer improves the signal to noise ratio of speech signals by coherently summing signals obtained from a linearly-spaced microphone array. Applications include speech recognition, acoustic echo cancellation, and automotive “hands-free” audio input systems.

 

The diagram above shows a communication system in which beamforming can be used. On the right side of the figure, a person is speaking to Person A and Person B, who are in the same room. Person A and Person B are communicating hands-free.

Notice that Person A’s speech has a direct path to the set of four microphones as well as an indirect path to these microphones. The beamforming algorithm causes the microphone gain to be maximum in the direction of Person A while she is speaking. By increasing the gain in that direction while reducing the gain in the direction of the reflective paths, we increase the signal-to-interferer ration, which reduces the reverb effect.

Similarly, but not show, this technique improves the signal to noise ratio because background noise tends to come from all directions, not just from the direction of the desired speech.

Although Adaptive Digital’s beamforming algorithm provides user-programmable microphone spacing, it is recommended that the microphones be closely spaced. The recommended spacing between adjacent microphones should be less than 10 centimeters. Furthermore, all microphones must be placed linearly and with equal spacing between them.

Care should be taken to ensure that analog circuitry and analog to digital converters for all microphones be designed with tight tolerances in order to provide for a minimum difference in gain, delay, frequency, and phase characteristics between the microphones.

Function API's​

API function call summary

beamform_ADT_init(…) Performs Beamforming initialization function

beamform_ADT_run(…) Performs Beamforming function

Translate »