Solutions for Bluetooth hands-free accessories
You are here: Solutions -> Bluetooth hands-free
Voice processing solutions
Alango has developed an extensive portfolio of digital signal processing technologies enhancing the voice communication quality of Bluetooth hands-free accessories (headsets, hands-free car kits, portable Bluetooth speakers and speakerphones). The technologies are seamlessly integrated into one Voice Communication Package (VCP) sharing memory and computational blocks to lower device power consumption and program memory. VCP has been ported and optimized for Kalimba DSP powering CSR BlueCore 5 Multimedia ICs.
In its maximal configuration VCP implementation running on BC5MM integrates:
- Adaptive dual microphone (headset and speakerphone applications)
- Acoustic echo cancellation (headset and speakerphone applications)
- Noise suppression (stationary and transient on microphone and speaker channels)
- Automatic microphone gain control and dynamic range compression (headset and speakerphone applications)
- Automatic speaker volume and equalization control (ambient noise dependent)
- 32 band static equalizer (microphone and speaker channels)
- (new!) Intelligent Boost Control
Alango VCP DSP package is fully compatible with CSR Q3 mono and stereo headsets. It implements exactly the same audio streaming and messaging mechanism and conventions as the standard CSR CVC DSP library. In practice, the VCP and CVC libraries can be interchanged rewriting just one line of the application code – the path to the library being integrated.
Music and entertainment solutions
Bluetooth stereo headsets are often in places where ambient noise level and spectrum change fast and in a wide range (street, metro, train, bus). Loud ambient noise often masks the audio content spoiling music experience of mobile users. Alango has developed a unique Automatic Volume and eQualization control technology for Music and Entertainment (AVQ-Me) enhancing music according to dynamically changing noisy conditions. This virtually masks the ambient noise delivering sound of perceptually equal loudness and timbre in varying noisy conditions.
AVQ-Me is now being implemented on CSR BlueCore 5 Multimedia IC. Scheduled time for its release is Q3 2008. Please, contact us if you want a delivery notification from Alango.
Voice Processing Technologies for Bluetooth accessories
Adaptive Dual Microphone (advanced technology)
Alango Adaptive Dual Microphone (ADM) technology is an innovative, proprietary technology that reduces all types of ambient noises surrounding a Bluetooth headset user. Adding the second microphone plus ADM technology to a Bluetooth headset enables efficient voice communication in much noisier conditions compared to a one-microphone headset. Using a Bluetooth headset in train stations, noisy restaurants, and discotheques is no longer a problem.
While having several competitors, ADM technology is characterized:
- Very efficient cancellation of all types of noises
- No degradation of user’s voice irrespective of the original signal to noise ratio
- Auto compensation for the microphones sensitivity mismatch
Automatic speaker volume and equalization (advanced technology)
Automatic Volume and eQualization control for Voice (AVQ-V) is an advanced technology that automatically adjusts the speaker sound according to the ambient noise. AVQ-V works in the same 32 frequency subbands as the other VCP technologies allowing precise equalization of the loudspeaker sound. AVQ-V monitors noise levels in each subband of the microphone signal and amplifies the corresponding bands in the loudspeaker signal accordingly when required.
AVQ-V technology is very helpful for Bluetooth headsets and hands-free car kits often used in circumstances where the ambient noise levels change fast and in a wide range. Selective amplification of loudspeaker signal frequency subbands masked by noise provides much more natural effect than simple automatic volume control keeping perceptually equal loudness in variable noisy conditions.
Acoustic Echo Cancellation
Alango Acoustic Echo Cancellation has been designed to provide full-duplex communication capabilities to Bluetooth devices. Alango algorithms support both headset and speakerphone types of applications.
Alango algorithms are distinguished by several important factors:
- Fast convergence. The typical convergence time to full duplex operation is about 200ms without initial echo.
- Robustness and convergence in double talk allowing two-way, echo free communication when acoustic conditions are changing during double talk (e.g. when the user covers the phone speaker by hand).
- Advanced, subband residual echo suppressor and comfort noise generator allowing smart, frequency selective echo suppression. This dramatically improves duplex performance of mobile devices where the initial echo is high and often non-linear.
- Robustness and good performance in the presence of high non-linear speaker distortions and mechanical speaker-microphone coupling. Alango technology allows direct specification of the level of loudspeaker distortion, its frequency region and dependence on the signal level.
- Large dynamic range allowing handling input signals of very different and fast changing levels.
Microphone noise suppression technology is important in almost any Bluetooth voice communication device. Alango VCP package additionally integrates speaker channel noise suppression improving user’s listening experience when the other party speaks from noisy conditions.
Alango single channel noise suppression technology is characterized by:
- Fast adaptation to changes in noise level. As fast as 100ms noise level adaptation time may be used without voice degradation. This allows canceling transient as well as stationary noises (passing cars, engine noise of an accelerating car, etc)
- High voice fidelity even for large noise levels
- No processing artifacts (musical noise etc)
- Wide dynamic range of input signals
Automatic microphone gain control
Microphone signal levels may be very different due to several reasons. Users have different voices and they talk different in different circumstances. People generally speak softly in quite environments while they may almost shout when ambient noise is high. In addition, according to a law of acoustics, the microphone signal amplitude in a free field is inversely proportional to the distance between the microphone and the sound source (user’s mouth). While it is relatively fixed for headsets, this distance may be very different for hands-free car kit or speakerphone applications.
As such, efficient automatic microphone gain control is one of the most important features of a good Bluetooth voice product. Most existing solutions employ strong Dynamic Range Compression (DRC) of loud signals. Simply, the louder the user speaks the more his voice signal amplitude is reduced (compressed). This causes flattening user’s voice signal level. However, since far sounds (noises) are not compressed, this causes significant worsening of the signal-to-noise ratio. This artificial degradation of signal-to-noise ratio caused by compression is one of the main reasons why distant noises often sound as loud as user’s voice however loud and close to the microphone he or she speaks.
Alango offers alternative solution by exploiting its fast and reliable AGC algorithm. Instead of compressing user’s voice when it is strong, Alango AGC technology digitally increases microphone channel gain when user’s voice is weak. When no user’s voice is detected by the Alango robust voice activity detector, no gain change is performed. In noisy places Bluetooth headset users tend to speak loudly which automatically sets the microphone channel gain to a small value thus reducing ambient noises while keeping user’s voice at the optimal level. Alternatively, in quiet environment people speak softly so their voice are amplified by the algorithm but, no ambient noise present anyway. In practice, a combination of AGC algorithm with a mild DRC algorithm (both from Alango) gives the best results.
In the hands-free or speakerphone applications the problem is slightly different. The relative noise level in the input signal may be large but the speed of voice level variations is generally lower. Again a combination of AGC and DRC algorithms gives excellent results with all people being heard well irrespective of the distance to the microphone.
go back to the Solution page.