|Alguien dijo ...|
|El arte es yo; la ciencia es nosotros,|
Claude Bernard (1813-1878).
|Zarlink Advanced Voice Processing Platform Improves Voice Quality in Hands-Free Communication Systems|
OTTAWA, CANADA - Zarlink Semiconductor (NYSE/TSX:ZL) today expanded its voice processing product portfolio with a highly integrated, flexible and programmable solution for hands-free communication systems. The ZL™38004 voice processing platform and embedded firmware delivers measurable sound quality improvements in high-noise environments. The integrated solution reduces development time and board space requirements..
The ZL38004 voice processing platform is a complete solution integrating dual-channel codecs, Zarlink voice processor engine for acoustic echo cancellation and noise reduction, and complete firmware support for hands-free car kits (ZLS38500) and speakerphones (ZLS38501). The new platform helps designers deal with complex noise situations, such as echo, background noise and double-talk that have traditionally plagued hands-free communication systems.
“Hands-free communication system manufacturers require voice processing solutions that can perform in complex noise environments,” said Andre Coucopoulos, product marketing manager, Voice Processing, Zarlink Semiconductor. “Designed specifically for high-end hands-free car kits and speakerphone systems, the ZL38004 chip with firmware delivers superior sound performance for end-users and reduced design complexity when compared with competing technologies.”
Zarlink will be demonstrating the ZL38004 voice processing platform at GLOBALCOMM 2006 (Booth#48088) with speakerphone manufacturer Multisuns. Zarlink’s new voice processing platform is designed into Multisuns’ ConferBASE phone system.
Voice performance in hands-free car kits
Manufacturers are developing high-performance hands-free car kits that deliver superior voice quality in severe noise situations. For example, hands-free car kits must deal with vehicle noise and acoustic echo created by direct coupling between the microphone and speaker and sound reverberation inside the car cabin.
Hands-free car kit manufacturers require voice processing solutions that deliver noise reduction, good double-talk and clipping compensation performance. Cost and board space savings are also a concern, particularly as hands-free car kits are deployed in higher volumes.
To achieve good noise reduction, the ZLS38500 firmware distinguishes between voice and noise signals. The firmware removes the noise from the total signal and preserves the integrity of the original speech signal.
Double-talk, when two people are speaking at the same time, commonly poses a problem for hands-free communication systems. Good double-talk performance relies on the interaction between the AEC (acoustic echo canceller) and NLP (non-linear processor). The ZLS38500 firmware tracks changes in the echo path and continuously converges during double-talk and the NLP cancels the imperfections of the AEC.
Under severe noise conditions, for example a car driving at over 100 km/h with the windows open, a good clipping compensation circuit is required. The ZLS38500 firmware prevents the signal from being clipped, allowing the AEC to converge to echo and the noise reduction circuit to identify and reduce background noise.
Complete solution for speakerphones
Speakerphone manufacturers consider sound quality as a key differentiator, and prefer an integrated solution that reduces design cost. Factors such as the quality of plastic enclosure used in the speakerphone design and the size of the room where the phone is used can also impact the quality of a voice processing solution.
The ZLS38501 firmware delivers exceptional full-duplex operation and good double-talk performance. Most voice processing solutions deliver only half-duplex operation, meaning voice signals pass in only one direction. During double-talk only the signal with the highest level is transmitted, resulting in interrupted conversation.
Most competing full-duplex solutions will also stop tracking the echo environment during double-talk. The algorithm then has to readapt when double-talk ceases, resulting in an audible burst of echo. In comparison, the ZL38004 voice processor algorithm is able to continuously converge and track changes in the echo path to support full-duplex operation.
Multisuns speakerphone demonstration at GLOBALCOMM 2006
Zarlink will be demonstrating the ZL38004 voice processing platform with speakerphone manufacturer Multisuns at GLOBALCOMM 2006. Multisuns’ ConferBASE phone systems are designed for multi-user teleconferencing. The ConferBASE phones carry all-direction speech tracking in 16 ms (millisecond) intervals with 20-30 degree resolution, and analyze audio waveform under relative loss of 20 dBm. With these capabilities, teleconference participants can easily recognize other callers.
Designed into the ConferBASE phones, the ZL38004 voice processing platform helps eliminate environmental noise in office situations, such as meeting room echo and air conditioner or cooling fan sounds.
“The ConferBASE teleconference system has been made more efficient and user-friendly,” said Bruce Sun, CEO, Multisuns Corporation. “Zarlink’s flexible voice processing platform provided a solution that can be fine-tuned to assist us in meeting the unique performance demands of our speakerphone systems.”
Viernes, 09 Junio, 2006 - 08:52