Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > Processors/DSPs

Acoustic echo cancellation and noise suppression libraries

Posted: 08 Mar 2005 ?? ?Print Version ?Bookmark and Share

Keywords:microchip technology? software library? dspic? digital signal controller? dsc?

Microchip Technology Inc. announced two new software libraries for its 16bit dsPIC Digital Signal Controllers (DSCs). The G.167-standard compliant dsPIC30F Acoustic Echo Cancellation Library offers improved performance for "speaker and microphone"-based products that either have no cancellation today, or can benefit from the reduced cost offered by Microchip's unique licensing structure.

The dsPIC30F Noise Suppression Library enables design engineers to gain a competitive edge for their microphone-based products with a methodology that goes well beyond standard filtering techniques for impressive intelligibility gains in environments subject to variable noise. These libraries can be used together to provide an ideal solution for applications with high levels of ambient noise and echo in compact environments.

The dsPIC30F's DSP instructions are used extensively in both libraries. Yet, despite the complexity of these libraries, the user interface is simplejust one library file and one header file. Example applications that can benefit from these advanced libraries include: hands-free cell phone kits, speaker phones, intercoms, emergency vehicles, teleconferencing systems, headsets and the front end of speech-recognition systems.

"These advanced libraries are an excellent way to quickly exploit the power of DSP without the development effort," said Sumit Mitra, VP of Microchip's Digital Signal Controller Division. "When superior product performance and time to market are both critical, these advanced libraries can really come to the rescue."

The dsPIC30F Acoustic Echo Cancellation Library supports full-duplex communications through two functions that are easily callable via a well-documented Application Programmer's Interface (API). One function is used for initialization and the second function removes the echo component from a 10ms block of sampled 16-bit speech data. These functions ensure that speaker-to-microphone induced echo is suppressed.

The dsPIC30F Noise Suppression Library also has two API-callable functions, initialization, and the removal of noise from a 10-millisecond block of sampled 16bit speech. This functionality provides microphone-based applications with the ability to ensure that only voice content is transmitted.

"Most people use complex analog filters to achieve noise suppression. However, detecting and suppressing a wide range of rapidly changing noise profiles requires something more," Mitra said. "With an evaluation license fee of $5, Microchip makes it easy for engineers to test drive our Noise Suppression Library or Echo Cancellation Library to determine whether the dsPIC DSC is right for their design."

With the dsPIC30F Acoustic Echo Cancellation Library, the received far-end speech samples are filtered using an adaptive Finite Impulse Response (FIR) filter. The coefficients of this filter are adapted using the Normalized Least Means Square (NLMS) algorithm, such that the filter closely models the acoustic path between the near-end speaker and the near-end microphone. A Non-Linear Processor (NLP) algorithm is used to eliminate residual echo. Voice activity detection and double talk detection algorithms are used to avoid updating filter coefficients when there is no far-end speech, and also when there is simultaneous speech from both ends of the communication link (double talk). This library is configurable for 16-, 32- or 64ms maximum echo delays (echo tail-lengths). The library is also compliant with the G.167 standard and has been tested for compliance with G.167 specifications for in-car applications.

In the dsPIC30F Noise Suppression Library, the noise suppression is primarily a frequency domain algorithm. The signal is sampled at 8kHz, and a Fast Fourier Transform (FFT) is performed on each 10ms block of data to analyze the frequency components of the signal.

Thereafter, a voice-activity-detection algorithm is used to determine whether the signal segment is speech or noise. The noise-suppression algorithm maintains a profile, which is updated each time a noise-only block is detected. Every frequency band of the input signal is scaled down in proportion to the noise in that frequency band, thereby causing a significant degree of noise suppression in the resultant signal. The algorithm adapts to changes in the nature and level of noise, and does not require a separate noise-reference input.

Availability and pricing

All licensing options for both libraries are immediately available. Among these options is a one-year evaluation license, which is unaltered from the libraries that are used for production and can be purchased online at The dsPIC30F Acoustic Echo Cancellation Evaluation Library license (SW300060-EVAL) is $5, and the dsPIC30F Noise Suppression Evaluation Library license (SW300040-EVAL) is also $5. Industry-unique, one-time-fee licensing options are also available for production, which runs counter to the per-unit royalty business model that is common for this class of library. Starting at $2,500 for 5,000-unit products, the license fee that Microchip charges is attractive to many customers since it is a fraction of the cost they would incur developing it themselves.

These one-time fees are based on the estimated lifetime volume of products derived from the design effort that uses the library.


Article Comments - Acoustic echo cancellation and noise...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top