The inert gas helium is used to dilute the oxygen that saturation divers breath so that, under the great pressures in which they live and work, excess oxygen will not be absorbed into their body tissues, causing illness or death. However, because the speed of sound in helium is three times that in air, sound reverberating in a HeO2 mixture produces higher frequencies--perhaps 2.7 times higher. Most of the frequencies in speech comes from reverberation of vocal cord sounds in vocal tract of the speaker. Thus, the speech of a saturation diver working at considerable depth sounds like this.

Working for the United States Navy, Oxford Signal Processing has developed a new generation of helium-speech unscrambler (HSU). This device translates helium speech back into normal sounding speech. The speech sample, above, becomes this.

The new HSU consists of three circuit cards (pictured on the right): an analog-to-digital card (ADC), a central processing card (CPU), and a digital-to-analog card (DAC). The CPU card employs two state-of-the-art digital signal processors (DSPs) that support more sophisticated algorithms than was previously possible. Besides the voice translation function, the new HSU also supports a noise reduction function. One DSP is dedicated to each function.

Since algorithm improvement is an ongoing process, provision was made for updating the software without removing the CPU card from the communication system. The operator connects a PC's parallel printer port to the DAC card. Th then powers up the communication system and the CPU card boots on data from the PC. It burns the new software into its flash memory and signals completion with five tone bursts. The operator powers down the system and removes the cable. When the system is powered up again, it runs on the new software. With this arrangement, new software can be sent to users on any portable medium, or even attached to email messages.

Oxford Signal Processing anticipates that the US Navy will release this technology soon for commercial applications. At that time, if there is sufficient demand, a stand-alone device will be designed around these circuit cards for use in communication systems that provide and accept standard audio line signals. Customized designs for specific communication systems are also possible.

Processor Card

Digital-to-Analog Card

(click for larger image)