PureSpeech®

• Reduces the effects of noise in the transmission path
• Dynamically maximises speech intelligibility and voice clarity through selective formant enhancement
• Send and Receive EQs enable improvements in user experience and enables poor microphone and speaker frequency responses to be optimised
• Echo Canceller reduces the echo due to acoustic coupling between speaker and microphone. PLC improves the Rx audio quality in the presence of bit and packet errors within the Bluetooth® link. PLC also reduces the effects caused by interferences such as cross-body, Wi-Fi® or RF degradation

Near-End Enhancement

• PureSpeech improves the ability to hear and understand the caller’s voice, regardless of the listener’s location. It monitors ambient noise while simultaneously analysing received speech
• As ambient noise rises, PureSpeech sharpens the key “formants” (or the high energy peaks in the speech spectra that are critical for intelligibility) in the spoken word while adjusting the gain applied to maximise voice intelligibility, even in the most difficult listening environments
• By selectively emphasising the speech components required for comprehension, intelligibility gains of 8 to 14dB can be realised based on word recognition scores
• In operation, the speech signal is separated into multiple bands and analysed for energy peaks by the near-end enhancement engine of PureSpeech. These peaks, or syllables, represent speech formants that are critical for intelligibility. In other words, these are the sounds that the human ear hears and brain processes to understand what is being said
• By normalising the formants based on a predefined weighting curve using psychoacoustic principles, and reducing the level of non-formant speech components, the PureSpeech process ensures that all formants cut through the ambient noise to become perceptible without significantly increasing the overall signal magnitude

Far-End Enhancement

• The noise suppression and echo cancellation components of PureSpeech benefit the call recipient by eliminating outbound noise and reducing echo due to acoustic coupling between the microphone and speaker
• As headset devices are frequently used in high noise environments, without some form of noise suppression, the ambient noise (other people speaking, noise from cars or trains, etc.) would be transmitted along with the desired speech signal. For this reason, it is highly desirable to suppress extraneous noise while maintaining the best possible speech quality
• PureSpeech carries out its noise suppression in the frequency domain, using the properties of speech to differentiate it from the noise and then remove the noise from the speech signal
• Automatic Gain Controls improve speech quality by normalising the outbound speech level. This is necessary since speech levels will differ between user and acoustic environments
• Acoustic echo cancellation is a critical function of any speech processing system for telecommunications. PureSpeech analyses speech in the receive path to identify and remove echo in the outbound signal path while maintaining full duplex operation
• Autocorrelation methods are employed to use speech in the receive path to identify echo in the outbound signal path. The use of a sub-band adaptive filter means that echo can be removed from the send path while maintaining full duplex operation