ADAPTIVE DIGITAL NOISE REDUCTION

ADT NOISE REDUCTION C54x , C55x , C64x

Voice Quality Enhancement Software

Adaptive Digital has two classes of noise reduction algorithms . They are distinguished by a number of characteristics.

Data sheet:.pdf

Adaptive Digital has two classes of noise reduction algorithms. They are distinguished by a number of characteristics:

CPU Utilization
Memory Utilization
Algorithmic Delay
Ability to cancel noise in the presence of signal. (SNR Improvement). If the algorithm does not cancel noise in the presence of signal, it only reduces noise when desired signal (speech) is not present as is the case between speech bursts.
Frequency Domain Based vs. Time Domain Based

We refer to the two classes of noise reduction algorithms as the standard version and the low complexity version. The table below compares the characteristics of the two versions.

Algorithm Version	CPU Utilization	Memory	Delay	SNR Improvement	Frequency / Time Based
Standard	High	High	Moderate	Yes	Frequency
Low Complexity	Low	Low	None	No	Time

The user can configure either version’s noise cancellation to be more or less aggressive. Less aggressive settings reduce noise by a moderate amount while more aggressive settings reduce noise by a greater degree.

One might wonder why a system would be configured with a less aggressive setting. There are a few reasons. One reason is that a very aggressive setting can result in loss of desired signal. Another reason is that, in the case of the low complexity version, an aggressive setting will cause an accentuated difference between the noise level during speech and the noise level when speech is not present. The transitions between speech and no speech may become more objectionable with the more aggressive setting.

APPLICATIONS

Standard noise reduction is an excellent choice for applications with high background noise environments such as mobile phones, speakerphones, hands-free phones, and defense communication systems. It is also useful when used in conjunction with voice recognition to increase accuracy.

Low Complexity noise reduction is an excellent choice for wireline telephony and conferencing.

FEATURES

User configurable
Enhances Voice Quality (improves Mean Opinion Score)
Functions are C-callable
Designed for multiple channel operation

AVAILABILITY

ADT Noise Reduction is available on the TMS320™ DSP Family

C54x™DSP, C55x™DSP, & C64x™DSP Generations

SPECIFICATIONS

Noise Reduction C54x

All Memory usage is given in units of 16-bit word.

Variant	MIPS (Peak)	Program Memory	Data Memory	Per Channel Data Memory
Standard	7.5	2272	155	750
Low Complexity	0.3	400	42	12

Last update: 05/01/2004

Noise Reduction C55x

All Memory usage is given in units of byte.

Variant	MIPS (Peak)	Program Memory	Data Memory	Per Channel Data Memory
Standard	7.5	4500	310	1500
Low Complexity	0.3	800	84	24

Last update: 05/01/2004

Noise Reduction C64x

All Memory usage is given in units of byte.

Variant	MIPS (Peak)	Program Memory	Data Memory	Per Channel Data Memory
Standard	7.4	10400	1384	1500
Low Complexity	0.2	2400	84	24

Last update: 05/01/2004

FUNCTION

NCAN_ADT_config( . . .) Configures Noise Canceller

NCAN_ADT_init( . . .) Initiates a channel of Noise Cancellation

NCAN_ADT_cancel( . . . ) Executes Noise Cancellation

APPLICATIONS

Standard noise reduction is an excellent choice for applications with high background noise environments such as mobile phones, speakerphones, hands-free phones, and defense communication systems. It is also useful when used in conjunction with voice recognition to increase accuracy.

Low Complexity noise reduction is an excellent choice for wireline telephony and conferencing.