This paper addresses the problem of increasing speech-in-noise intelligibility under the constraint of energy preservation. Two recently proposed algorithms which have been shown to be very successful in this problem according to two large formal listening tests are reviewed and a hybrid system which combines the properties of the two methods is suggested. The first technique, which is a frequency domain approach, is re-implemented providing clarifications on its energy reallocation strategy. Based on objective measures well correlated with human perception, we show that our implementation performs similarly to the original approach. Moreover, this is combined with a dynamic range compression algorithm from the second method to allow reallocation of energy over time as well. Experiments with speech shaped noise (SSN) and competing speaker (CS) noise maskers at various SNRs indicate that the hybrid system outperforms the individual algorithms in terms of intelligibility scores.