Some qualitative data on the nature of the speech distortion caused by respiration of a helium-oxygen atmosphere are presented. Previous corrective-processing techniques are discussed, and an advanced technique that overcomes their disadvantages described. The experimental system is based upon time-domain speech-processing methods and is capable of retaining the natural pitch rate of the speaker's voice while correcting the formant shifts caused by the helium-rich atmosphere. This system uses digital-processing methods and is capable of operating on continuous speech in on-line applications. It promises a practical solution to the problem of providing normal voice communication in a helium environment.